Rail Punctuality in Britain Germany and Switzerland

Executive Summary

• Headlines measure different things and cannot be ranked. Great Britain's 84.8% in 2024-25 counts recorded station stops within 3 minutes ^[2], Germany's 62.5% long-distance figure counts Fernverkehr arrivals within 5:59 minutes ^[37], and Switzerland's 93.2% counts SBB trains arriving with less than 3 minutes' delay at destination/station ^[31]. ORR explicitly warns direct comparisons are limited by measurement differences ^[14][17], a warning echoed for Europe more broadly ^[24].
• Whole-network averages hide a long-distance gap, especially in Germany. DB rail passenger transport in Germany was 89.5% in 2024 ^[38][41] while DB Fernverkehr operational punctuality was only 62.5% ^[37], down from 64.0% in 2023 ^[37] to a historic low ^[23]. In Great Britain the stricter all-stops On Time (<1 minute) was 67.0% for the 12 months to December 2024, while terminus-based PPM was 85.2% for the same period ^[14].
• Cancellations are treated differently and change what the denominator means. Published punctuality rates in Germany exclude cancellations, recorded separately ^[11][28][30], while GB punctuality deliberately excludes failed stops because they are captured in reliability measures ^[29]. GB's official cancellations score was 4.1% in 2024-25, a record high ^[2], but the timetable is finalised at 22:00 the previous evening and earlier removals are excluded ^[12][14][29]; including resource-availability P*-coded pre-cancellations raised Q4 2024 from 5.1% to 5.7% ^[14].
• Coverage, period and completeness differ. GB covers mainline passenger trains in Great Britain ^[3], surface rail only excluding underground, light rail and tram ^[2], for financial year 2024-25 ^[2] with about 96% of stops recorded and no estimate for the remainder ^[14]. DB reports calendar-year shares of punctual stops among intermediate and terminal stops for >800,000 monthly runs ^[35], and SBB's 93.2% is an operator headline for arrival at destination ^[31], distinct from BAV regional quality scores and freight customer punctuality of 88.0% ^[25].
• Recommendation: present definitions, thresholds, denominators and cancellations side-by-side with caveats beside each figure — never as a single league table. Valid comparison requires preserving units, stating On Time versus Time to 3 versus <6 minutes versus <3 minutes, and showing official versus adjusted cancellations together.

1. How punctuality is defined: thresholds, measuring points and timetables

Main finding: the three systems use incompatible thresholds, measuring points and timetable baselines, so percentages are not interchangeable.

1.1 Great Britain — all-stops, minute-by-minute measurement against the plan of the day

Great Britain measures every recorded stop against the public timetable, with a family of thresholds rather than one number.

• Since 1 April 2019 the industry introduced a more detailed set of measures to track timeliness from station to station and minute to minute, reflecting that good performance must be delivered throughout the whole journey ^[7]. The definitions document confirms the move from terminus-only to whole-journey measurement from the start of Control Period 6, replacing the 2014 PPM definition ^[29].
• On Time means less than 1 minute late — time to 59 seconds — versus the public timetable (GBTT) in TRUST, with early trains counted as on time ^[29][14]. Time to 3 and Time to 15 are cumulative percentages of recorded stops arrived at early or less than 3 and 15 minutes after scheduled time ^[40][17]; technically less than x minutes, so Time to 3 is within 2:59 and Time to 15 within 14:59 ^[29].
• The legacy Public Performance Measure (PPM) remains terminus-based: proportion arriving at final destination early or less than 5 minutes after schedule for London and South East, Regional and Scotland operators, or less than 10 minutes for Long Distance operators ^[14][17], with Long Distance at 10 minutes and all other sectors at 5 minutes ^[29].
• The lateness of each train is assessed at origin, at arrival at each intermediate station and at termination, with each recorded station stop given equal weighting ^[29]. A Recorded Station Stop requires both a planned GBTT time and an actual TRUST recorded time, and TRUST does not currently capture all platforms and stations ^[29].
• Punctuality is measured against the public-facing passenger timetable (GBTT), with early trains included ^[29]. All performance metrics unless specified otherwise are measured against the Applicable Timetable as advertised to the public at 22:00 the day before operation ^[29][12]. Trains cancelled prior to 22:00 the night before are excluded ^[29].

1.2 Germany — operational train punctuality (<6 minutes) plus passenger-weighted traveller punctuality (<15 minutes)

Germany publishes a train-based operational rate with a generous threshold alongside a passenger-weighted rate that includes cancellations and missed connections.

• DB measures punctuality by continuously comparing actual versus scheduled arrival for each operated train/bus run and aggregating on-time plus arrivals delayed up to a defined maximum into the punctuality rate ^[41]. The reference is arrival rather than departure ^[35].
• Operational punctuality (betriebliche Pünktlichkeit) is punctual arrival plus a maximum of 5:59 minutes, i.e. exceeding scheduled arrival by less than 6 minutes ^[35]. A train enters the delay statistics from 6 minutes ^[30][28]. DB notes the less-than-6-minute stop rule follows practice of many other European railways ^[35]. In Baden-Württemberg the contrast is explicit: DB counts punctual as under 5 minutes 59 seconds, while most state networks require at most 3 minutes 59 seconds ^[6].
• Reported operational percentages are the share of punctual stops among all intermediate and terminal stops (Anteil pünktlicher Halte an allen Unterwegs- und Endhalten) ^[35]. DB Fernverkehr defines operational punctuality as share of arrivals at scheduled stops with deviation < +6 Minuten ^[37].
• DB Fernverkehr additionally determines traveller punctuality (Reisendenpünktlichkeit) as punctual arrival of passengers at the booked destination station ^[35], defined as share of customers with < +15 Minuten delay at the end of the planned long-distance chain ^[37], or plus a maximum of 14:59 minutes versus planned arrival ^[35]. The threshold is noted as matching long-distance bus and air practice ^[35].
• Traveller punctuality explicitly takes into account working or missed connections, cancellations and replacement trains, timetable changes and alternative connections ^[35]. It counts the share of travellers deemed punctual at destination, not the share of punctual trains ^[36], and covers the complete journey experience including all delays, cancellations, replacement and alternative options and whether all connections are made ^[36][37]. Occupancy matters: one ICE may carry 90 and another 900 passengers, and a delayed train can still yield an on-time final arrival if the connection is caught ^[36].

1.3 Switzerland — arrival-based <3-minute rule, with train, connection and regional-quality layers

Switzerland uses the strictest headline train threshold, but SBB distinguishes train arrival, connection maintenance and customer arrival.

• In Switzerland a train counts as punctual if it arrives with less than 3 minutes' delay ^[31], measured on arrival at the station ^[31]. The headline refers to arrival at destination (Bestimmungsort) ^[31]. A journey counts as punctual when it has less than 3 minutes' delay ^[34].
• Unlike most European railways, SBB has measured customer punctuality rather than only train punctuality ^[20]. A journey is punctual when the passenger arrives at the destination with less than 3 minutes' delay and all connections could be maintained ^[20]. Most other railways define a train as punctual with less than 5 minutes' delay without considering connections ^[20]; in Germany a train with 5 minutes 59 seconds is still punctual ^[20].
• SBB reports connection punctuality (Anschlusspünktlichkeit) as a separate indicator alongside train punctuality ^[25]. In the regional quality system, digital punctuality data (DPM) uses the SBB CUS customer information system as source ^[16], and quality is monitored through DPM plus test-customer surveys (MSS) ^[34].

2. Official 2024 headline figures: whole-network versus long-distance

Main finding: 2024 was a record high for SBB, a historic low for DB long-distance, and a slight deterioration in Great Britain — but periods and scopes differ.

2.1 Great Britain — financial-year Time to 3 plus calendar-year On Time and PPM

Great Britain's headline depends on which threshold and which 12 months are quoted.

Rail Trends 2025, covering surface rail only excluding underground, light rail and tram for financial year 2024-25 ^[2], reports 84.8% of trains arrived within 3 minutes (early or less than 3 minutes after scheduled time) ^[2]. That was a decrease of 0.4pp on the previous year and an increase of 0.8pp on 2019-20, sourced to ORR table 3133 ^[2]. The chart is labelled proportion arriving at recorded station stops within 3 minutes ^[2].

For context, punctuality was stable around 84–85% before the pandemic, then improved to 92.9% in FYE March 2021 when 92.9% arrived within 3 minutes, driven by reduced services and passenger numbers during COVID-19 ^[2].

Calendar-year detail for 2024 on the stricter and terminus measures:

• In Q4 2024 (1 October to 31 December 2024), 62.1% of recorded stops (12.7m out of 20.4m) were arrived at On Time ^[14]. For the 12 months to December 2024 (1 January to 31 December 2024), 67.0% (54.8m out of 81.7m) were On Time ^[14].
• Using PPM, 81.4% of trains were punctual at final destination in Q4 2024, and 85.2% for the 12 months to December 2024 ^[14].
• Earlier in the year, April–June 2024 saw 70.1% On Time (14.4m out of 20.5m) and PPM of 87.4% ^[17]. For the 12 months to June 2024, On Time was 67.5% (53.6m out of 79.4m) and PPM 85.6% ^[17].

On Time is defined as early or less than 1 minute after scheduled time, with early classified as on time ^[14][17]. PPM thresholds are 5 minutes for London and South East, Regional, Scotland and Heathrow Express versus 10 minutes for Long Distance, Hull Trains, Grand Central and Lumo in the Q4 release wording ^[14].

Caveat beside the figures: up to around 96% of all station stops are currently recorded, with no estimates for the circa 4% not recorded ^[14][17]. Periodic operational data are supplied by Network Rail on a 4-weekly basis ^[3], and tables can take up to 15 working days after period end to update pending validation ^[3].

2.2 Germany — high whole-network rate, very low Fernverkehr rate

Germany's whole-network average looks strong only because regional and S-Bahn volumes dominate; long-distance operational punctuality collapsed.

DB calendar-year 2024 rates, all using the <6-minute stop logic unless labelled traveller:

• DB rail passenger transport in Germany (DB-Schienenpersonenverkehr): 89.5% in 2024, down from 90.3% in 2023 and 91.0% in 2022 ^[38][41].
• DB Group rail in Germany: 89.4% in 2024, down from 90.1% and 90.9% ^[38][41].
• All rail traffic in Germany including external operators: 88.1% in 2024 ^[39].
• DB Regio: 90.7% in 2024, down from 91.4% and 92.2%, reported on an adjusted third-weighted basis for AC S-Bahn, DC S-Bahn and non-S-Bahn regional rail ^[41].
• DB Fernverkehr operational: 62.5% in 2024, down from 64.0% in 2023 and 65.2% in 2022 ^[41][37], described as a historic low ^[23]. The target of 71.5% operational punctuality was clearly missed with the fall to 62.5% ^[37].
• DB Fernverkehr traveller punctuality: 67.4% in 2024, down from 68.9% in 2023 and 69.3% in 2022 ^[41][37].

Supporting detail from DB Fernverkehr 2024: 4.0% trains cancelled versus 2.7% in 2023; average total delay per stop 10.2 minutes versus 9.6; 46.4% trains late at end of run versus 43.3% ^[37]. Around 61% of Fernverkehr trains were affected by at least one construction measure versus around 62% prior, and load-related delays especially train-sequence conflicts were about 60% of all delay events and the main cause ^[37]. DB said 2024 development was shaped above all by poor infrastructure condition ^[23], describing the network as dilapidated and overloaded ^[23], with GDL strikes in Q1 also negative ^[23]. Longer trend: long-distance was still 74.4% in 2015 ^[32].

Caveat beside the figures ^[35][6]:

• DB statistics cover more than 800,000 passenger-train runs per month, with more than 20,000 long-distance and around 780,000 local runs including all S-Bahn, with all traffic stops flowing into the survey ^[35].
• Traveller punctuality evaluated on journeys of on average more than 10m monthly long-distance passengers ^[35].
• Threshold sensitivity in Baden-Württemberg: in 2024 about 80% punctual under strict 3:59 definition, but only about 88% even under DB less strict under-5:59 definition ^[6].

2.3 Switzerland — SBB record on train punctuality, with regional and freight splits

SBB's 2024 passenger headline was its best on record, but it is an operator measure, not a whole-country all-operator average.

• In 2024, 93.2% of all SBB trains were on time, a record ^[25][31], reaching destination on time ^[31]. That was 0.7pp more than in 2023 ^[31].
• Regional splits: Region Mitte 93.71% and Region Ost 93.65% were the highest reported ^[25]; Western Switzerland 91.9%, up 2.7pp ^[31]; Ticino 92.6%, an all-time high up 2.1pp ^[31]. Nearly all regions improved, with West Switzerland and Ticino contributing significantly ^[25].
• Connection punctuality was 98.7% in 2024, unchanged from prior year ^[25]. Freight customer punctuality, distinct from the passenger headline, fell 1.8pp to 88.0% ^[25][31].
• Disruption notes: heavy snowfall on 22 November 2024 caused a punctuality negative record ^[25], illustrating vulnerability to extreme weather alongside other short severe events ^[25][31]. In German-speaking Switzerland long-lasting construction and poor quality of incoming international traffic hurt punctuality, with the St. Gallen St. Fiden landslide further reducing robustness ^[25].

Caveat beside the figures ^[34][16][20]:

• BAV regional quality differs in scope and units: regional rail 95.37 points in 2025 vs 94.77 in 2024, buses 90.26 vs 89.78, with express/RE lowest at 91.71 but strongest improvement ^[34]; 13 of 23 railway companies above minimum ^[34].
• Minimum train standard 93.00 points with 3.00 tolerance and 96.00 acceptance ^[16]; assessments cover only RPV lines, not local-traffic lines ^[16]; detailed operator table (e.g. SBB 95.14, BLS 96.62, SOB 97.65, RhB 90.87, NStCM 85.73 ^[16]) is QMS RPV 2025 dated May 2026, not 2024 figures ^[16].
• Older context: in 2014, 87.7% of SBB travellers reached destination at most 3 minutes late, up 0.2pp ^[20], while train-based 5-minute comparison gave 96.8% SBB vs 96.7% Austria and 94.5% DB ^[20].

3. How cancellations, skipped stops and coverage change the denominator

Main finding: who counts a cancelled or short-formed train — and which timetable is used — moves the result more than a few tenths of a point suggest.

3.1 Great Britain — weighted cancellations against the 22:00 plan, plus hidden pre-cancellations

The official score understates passenger-facing cancellation when trains are removed before 22:00.

• The cancellations measure is a weighted score counting full cancellations as one and part cancellations as half ^[2][3][12]. A train is cancelled if it failed to stop at one or more planned stops, always including origin and terminus ^[29]. Full is <50% of planned mileage including not running at all; part is ≥50% to <100% or 100% mileage but missing ≥1 planned stop ^[29]. Score is sum of weights as % of trains planned in the Applicable Timetable ^[29]. Trains arriving 120+ minutes late at destination do not count towards the cancellation score ^[29].
• The timetable is finalised at 22:00 the previous evening; trains removed before then are excluded ^[12][14][29]. The official score measures disruption against the timetable operating on the day ^[12]. Operational data on resource-availability pre-cancellations are supplied by Network Rail and operators on a 4-weekly basis ^[12], with table updates up to 15 working days after period end ^[12].
• Impact: official 4.1% in 2024-25, up 0.3pp to a record high as planned trains recovered post-pandemic ^[2]. In Q4 2024 official was 5.1% but adjusted including P*-coded pre-cancellations was 5.7%, 0.6pp higher ^[14]. In April–June 2024 official was 3.5% and adjusted 4.0%, 0.5pp higher ^[17].
• P*-coded pre-cancellations are changes due to non-availability of staff or rolling stock included in a revised timetable ^[12] that may not appear in operators' scores ^[12][2]. They can be confirmed as late as 10pm previous evening ^[21] and are excluded from the baseline timetables performance is measured against ^[21]. In one autumn example TransPennine Express was actually cancelling 20–30% but required to report only 5.6–11.8% on-the-day, including a Wednesday with 45% of 333 timetabled trains cancelled, 100 pre-cancelled night before ^[21]. ORR acknowledged a significant gap to passenger experience ^[21], now requires specific resource-availability data and publishes it alongside official statistics ^[21]. From January 2023 ORR publishes periodic P-coded data capturing shortages that may not appear in scores ^[2].
• Interaction with punctuality: CP6 punctuality deliberately excludes planned stops where the train failed to stop because those are captured in reliability ^[29]. But other train-based terminus measures re-include them: PPM counts any Total or Other Cancellation as not punctual over trains planned ^[29]; CaSL counts part/full plus ≥30 min late at final destination over trains planned ^[29]; Right Time requires <59 seconds at destination having called at all stops, with full/part as failures ^[29]. Average Passenger Lateness estimates how late every passenger reaches destination using Schedule 8 Performance Minutes weighted by alightings ^[29].
• Passenger expectations differ from rules: 57% felt any missed stop should count as full not partial, and the same share felt trains removed a week in advance should still count as cancellations ^[5]. When forced to choose, around two-thirds felt punctuality mattered more than fewer cancellations ^[5], while 69% felt within-1-minute punctual was fair and 75% felt within-3-minutes was fair ^[5].

3.2 Germany — operational excludes, traveller includes; partial terminations are contested

A heavily delayed train that is turned early disappears from operational punctuality but still harms traveller punctuality.

• Cancelled trains (Ausfälle/Zugausfälle) are excluded from operational punctuality ^[28][30] but included in traveller punctuality ^[28][30][36]. Published rates exclude cancellations, recorded separately as partially or fully cancelled runs with data collected directly from infrastructure managers ^[11].
• DB says cancelling a heavily delayed train is operationally sensible when it limits effects on others and maintains labour-law crew requirements ^[26], and that a heavily delayed train burdens the overcrowded network so partial sections may sensibly fail ^[26], provided alternatives exist ^[26]. Reporting cited an internal message for ICE 616 Munich–Hamburg terminated early in Cologne as "cancelled from Cologne to improve the statistics" and a similar justification 11 September ^[26]. Metrics carry weight for management because they feed internal evaluation of divisions ^[26]. Since 2024 the federally owned InfraGO bundles track and stations, with federal funds tied to quality goals such as punctuality ^[26].
• Coverage nuance: reliability in Baden-Württemberg measures how many planned train-km were actually operated, with unplanned cancellation worsening the rate ^[6]; it fell from over 98% in 2014 to about under 96% in 2024 ^[6]. Traveller punctuality's inclusion of cancellations is why it can exceed operational in a bad year for terminations yet still fall: 67.4% traveller versus 62.5% operational in 2024 ^[37].

3.3 Network coverage — mainline versus operator versus regional system

Each headline covers a different network, so mix effects dominate.

• Great Britain statistics cover passenger trains operating on the mainline network ^[3], quarterly and periodic (4-weekly) on punctuality, reliability and causes ^[3]. Official coverage includes all franchised operators plus Hull Trains, Grand Central and Heathrow Express, even if wholly off Network Rail network such as Dalston Junction to New Cross ^[29]. Scotland is funding-based not geographic, excluding some trains in Scotland like CrossCountry Aberdeen–Edinburgh and including some wholly outside Scotland; England and Wales is all Applicable Timetable services not defined as Scotland ^[29].
• Germany's all-rail 88.1% includes external and internal operators ^[39], distinct from DB passenger 89.5% ^[38] and DB Regio 90.7% third-weighted ^[41]. Long-distance covers ICE/IC measured by punctual arrival at destination ^[30][32]; September detail illustrates split: Fernverkehr 61.9% versus prior-year 67.9%, ICE/IC only 55.3% down ~7pp, regional 87.2% versus 89.2% prior September ^[30]. Since 2021 the bwegt ranking separately measures regional SPNV quality ^[6].
• Switzerland's SBB headline is operator-specific, while BAV QMS RPV covers subsidised regional transport with test customers regularly checking quality and publishing results ^[8], using ~60 part-time test customers measuring punctuality, cleanliness, condition and information ^[8] plus company punctuality data flowing into QMS ^[8], recorded centrally and evaluated ^[8] as a steering instrument for Confederation and cantons with improvements via target/service agreements ^[8].

4. Why headlines cannot be ranked and what uncertainties remain

Main finding: a cross-country bar chart would be precise but wrong; the evidence providers themselves prohibit it.

• ORR states direct comparisons with other countries are limited due to differences in measurement ^[14], repeated as limited direct comparisons with other European countries ^[17]. The European Passengers' Federation confirms detailed cross-country comparisons are difficult because definitions differ ^[24].
• Six concrete incompatibilities from the evidence:
  1. Thresholds: <1 min / <3 min / 5-or-10 min in GB ^[14][29] versus <6 min operational and <15 min traveller in Germany ^[35][37] versus <3 min arrival in Switzerland ^[31][34]. Rounding differs, whether at half minutes or including all seconds to next minute ^[24].
  2. Measuring points: several places along the line versus only arrival station ^[24]. GB weights every recorded stop equally ^[29]; DB shares punctual stops among intermediate + terminal ^[35]; SBB headline is arrival at destination/station ^[31] with connections separate ^[25].
  3. Train versus passenger: train arrival punctuality explains only ~80–90% of variation in passenger punctuality and is an imperfect proxy ^[1]. Switzerland and the Netherlands measure proportion of delayed passengers, not just punctual trains ^[24]. Germany's traveller rate weights by occupancy and connections ^[35][36].
  4. Cancellations denominator: unclear whether statistics cover only performed departures or all planned including cancelled ^[24]. GB excludes pre-22:00 removals ^[29][12]; Germany excludes cancellations from operational but includes in traveller ^[28][30]; GB excludes failed stops from punctuality but captures in reliability ^[29].
  5. Train-type harmonisation: no harmonized definition of long-distance versus regional ^[24]; Switzerland does not report separate long- versus medium/short statistics ^[24] in that European comparison, while GB splits Long Distance PPM at 10 min ^[29] and DB splits Fernverkehr versus Regio ^[30].
  6. Timetable and operations: timetables include buffer over bare running time to improve robustness, so padding affects measured punctuality ^[1]; higher network intensity is associated with reduced punctuality ^[1]; dense tightly scheduled networks propagate small delays rapidly ^[1]; more stops negatively influences punctuality ^[1]; service type (Sprinter vs Intercity) and rolling stock matter ^[1]. Switzerland and Netherlands have very high capacity utilisation yet very high punctuality ^[24], while Baltic high punctuality may reflect very low volumes and lower propagation risk ^[24].
• Additional uncertainty: delay data are heavily right-skewed so median-optimized predictions sit below the mean because rare massive delays pull the mean up ^[1]; MAE/RMSE are scale-dependent in original units and cannot be compared directly across delay magnitudes ^[1]; conclusions are highly sensitive to aggregation (mean favours Random Forest, median favours XGBoost) ^[1]; weather imputation and single-split training add uncertainty and should be exploratory ^[1]; disruption records require >5 min increases yet still miss cases ^[1]; peak-only, weekday-only coverage differs from whole-network ^[1]. These points come from a Netherlands modelling study, so transferability is illustrative, not directly calibrated to GB/Germany/Switzerland.
• Presentation rule: because thresholds, points, denominators and periods differ and there is no standardized evaluation framework and consistent benchmarks limiting comparability ^[1], the valid visual is not a ranked league table but small, like-for-like panels (e.g. GB Time to 3 over time; DB Fernverkehr operational versus traveller over time; SBB train versus connection over time), each with its threshold, denominator and period labelled beside the figure and cancellations shown separately. No such cross-country chart is provided here because it would violate the providers' own comparability warnings.

5. Regulatory reporting rules and how to present valid figures with caveats

Main finding: each country has a different legal publisher, cadence and assurance model — accreditation in GB, operator plus infrastructure-manager reporting in Germany under EU passenger-rights rules, and BAV test-customer plus company-data quality monitoring in Switzerland.

5.1 Great Britain — accredited official statistics from Network Rail operations data

• Passenger Rail Performance (On Time and Cancellations) are Accredited Official Statistics from ORR ^[2], with practice regulated by the Office for Statistics Regulation ^[3]. Releases, tables including table 3133, dashboard, methodology and quality information are on the ORR data portal ^[2][3]. Network Rail supplies periodic operational data on a 4-weekly basis ^[3]; tables publish after loading/validation and resolving queries, up to 15 working days after period end ^[3]. Data are shown by financial-year quarter Q1 April–June through Q4 January–March ^[3]. Since CP6 the industry suite is described as the most transparent in Europe ^[4], with On Time and cancellations better reflecting passenger experience and infrastructure operation ^[4]. Network Rail must account to customers and regulators ^[4], after a November 2018 Provisional Order on performance planning and recovery ^[4]. From April 2026 measurement changes will reflect passenger views gathered with Transport Focus, Network Rail and ORR ^[5]. There were 8.3m delay compensation claims closed in 2024-25, up 9% ^[2].

Valid GB presentation: always state financial versus calendar period, threshold (On Time vs Time to 3 vs PPM with 5/10 min split), denominator (recorded stops vs trains at terminus with counts like 12.7m/20.4m ^[14]), recording completeness (~96% ^[14]), timetable baseline (22:00 day before ^[29]), and official cancellations alongside adjusted P-coded rate (e.g. 5.1% vs 5.7% ^[14]).

5.2 Germany — EU service-quality reports, EBA enforcement and Bundesnetzagentur market data

• Railway service-quality reports are made available by undertakings under Article 29(2) of Regulation (EU) 2021/782, previously Article 28(2) of Regulation 1371/2007 now repealed ^[18]. Undertakings must monitor own performance ^[18] and publish on their website by 30 June 2023 and every two years thereafter, also on the Agency website ^[18]. The Agency does not verify content, own reports or assume responsibility ^[18].
• The national enforcement body is Unit 54 of the Federal Railway Office (EBA), subordinate to the Federal Ministry for Transport, seated in Bonn with staff in Hamburg and Karlsruhe ^[13]. It can issue binding remedy orders under Sections 5/5a AEG enforced with fines up to 500,000 Euro subject to judicial review ^[13], working via complaint handling plus proactive supervision under Article 32(1) ^[13]. Exemptions exist for historic/tourist and city/suburban/regional services ^[13]; operators must establish emergency plans for major disruptions ^[13]. Supervision mixes unprompted checks, reactive investigations, on-site inspections, audits and focus checks ^[13]. In 2023 audits of service-quality publication found no infringements ^[13]. Complaints in 2023–2024 centred on delays, missed connections and cancellations especially Article 18 onward travel, plus Article 19(10) compensation refusals ^[13]. Construction markedly increased in 2024 with time-limited closures shifted to road ^[13], using replacement buses for construction, disruption, accidents or closures ^[13].
• Bundesnetzagentur market investigation punctuality excludes cancellations, recorded separately as partial/complete failures with data collected directly from infrastructure managers ^[11].

Valid German presentation: always pair operational (<6 min, stops-based, cancellations excluded ^[35][28]) with traveller (<15 min, passenger-based, cancellations and connections included ^[35][36]), state Fernverkehr versus Regio versus all-rail scope ^[30][41][39], and show cancellations (e.g. Fernverkehr 4.0% ^[37]) separately rather than folding into operational.

5.3 Switzerland — BAV quality measurement for subsidised regional transport plus SBB operational reporting

• BAV has test customers regularly check quality in subsidised transport and publishes results ^[8], using QMS to ensure funders and passengers receive high-quality RPV services ^[8]. Data collected are centrally recorded and evaluated ^[8] and serve Confederation and cantons as steering instrument, with improvements via target/service agreements ^[8].
• QMS RPV 2025 results are dated May 2026, not 2024 ^[16], so they cannot substitute for 2024 headlines. Minimum train standard 93.00 points, tolerance 3.00, acceptance 96.00; bus 89.00/3.00/92.00 ^[16]. Several operators supply insufficient or no DPM data (e.g. train: FLP, LEB, TPF, TRAVYS, TRN, SBB-D insufficient, FART none ^[16]), limiting whole-system inference.

Valid Swiss presentation: label SBB train punctuality (<3 min arrival at destination ^[31]) separately from connection punctuality ^[25], freight customer punctuality ^[25] and BAV QMS points for RPV only ^[16][34], noting joint federal-cantonal funding around CHF 2bn per year ^[34].

Limitations / Open Questions

• Period mismatch cannot be resolved from the evidence provided. GB headline is financial year 2024-25 ^[2] with calendar-year cuts to December 2024 ^[14], while DB and SBB headlines are calendar 2024 ^[37][31]. No calendar-2024 GB Time to 3 annual average is given in the excerpts.
• Swiss denominator and cancellation treatment are thin. Evidence confirms <3-minute arrival and CUS sourcing for DPM ^[31][16] but does not state SBB's cancellation inclusion, timetable-freeze rule, recording completeness or national all-operator share. Customer punctuality for 2024 is not provided; only 2014 customer context (87.7% within 3 min ^[20]) and 2024 connection punctuality (98.7% ^[25]) are available.
• German whole-network cancellation rate is missing. Only Fernverkehr cancellations (4.0% 2024 ^[37]) and the principle that published rates exclude cancellations ^[11] are evidenced. Early-handling and exact frozen-timetable rule for DB operational are not specified in the excerpts.
• Recording completeness is not comparable. GB quantifies ~96% recorded with no estimate for ~4% ^[14]; DB quantifies volume (>800k runs/month ^[35]) but not completeness; SBB completeness is not quantified. Delay-responsibility rounding can sum to 99.9% as a system-generated figure ^[7], and official statistics include only On Time to Time-to-30 plus Early and ≥30 min ^[29], so tail behaviour is truncated.
• Mix and intervention effects are unadjusted. Buffer time ^[1], stop counts ^[1], intensity ^[1], construction exposure (61% of Fernverkehr trains affected ^[37]; southern Germany flood June 2024, third-party and official interventions, GDL strikes ^[41]; overloaded core network with every second facility needing renewal ^[27]), and SBB weather/landslide/international inflow effects ^[25][31] all shift rates but are not standardised. Investment level and demand are suggested as drivers of overall punctuality ^[24], and lorry taxes covering two-thirds of ~EUR 2bn Swiss maintenance ^[24] plus Taktfahrplan short interchanges since the 1980s ^[24] are cited as structural context, but per-capita investment comparison reflects only a given year so quick conclusions are difficult ^[24].
• Revisions and provisional status. ORR notes data are loaded/validated with queries resolved before publication ^[3]; Network Rail notes unless specified data are initial indications subject to change ^[4]; latest ORR data are typically provisional as operator cancellation updates can change figures ^[40]. No 2024-specific revision log is in evidence.
• No league table should be inferred. Given different thresholds, points, passenger-versus-train units, cancellation baselines and coverages, plus explicit warnings ^[14][17][24] and lack of harmonized train types ^[24], any single ranking would mislead. If evidence is thin on a comparison, the report states so rather than interpolating.

Sources

^[1] Predicting Train Delays and Punctuality Using Machine ... — https://studenttheses.uu.nl/server/api/core/bitstreams/491205d1-97b4-43a0-9c55-32394b4aac10/content · academic ^[2] Rail Trends 2025 — https://www.gov.uk/government/statistics/rail-trends-2025/rail-trends-2025 · government ^[3] Passenger rail performance | ORR Data Portal — https://dataportal.orr.gov.uk/statistics/performance/passenger-rail-performance/ · government ^[4] Performance - Network Rail — https://www.networkrail.co.uk/who-we-are/how-we-work/performance/ · government ^[5] Transport User Voice January 2025 - What counts as a late train? - Transport Focus — https://www.transportfocus.org.uk/news/transport-user-voice-january-2025-measuring-train-performance/ · government ^[6] Wie pünktlich und zuverlässig ist die Bahn in Baden-Württemberg? — https://vm.baden-wuerttemberg.de/de/mobilitaet-verkehr/bahn-und-bus/mehr-qualitaet-im-oepnv/qualitaet-puenktlichkeit-und-zuverlaessigkeit · government ^[7] Railway Performance - Network Rail — https://www.networkrail.co.uk/who-we-are/how-we-work/performance/railway-performance/ · government ^[8] Qualitätsmesssystem — https://www.bav.admin.ch/de/qualitaetsmesssystem-rpv · government ^[11] Marktuntersuchung Eisenbahnen 2024 — https://data.bundesnetzagentur.de/Bundesnetzagentur/SharedDocs/Downloads/DE/Sachgebiete/Eisenbahn/Unternehmen_Institutionen/Veroeffentlichungen/Marktuntersuchungen/MarktuntersuchungEisenbahnen/MarktuntersuchungEisenbahnen2024.pdf · government ^[12] coded cancellations | ORR Data Portal — https://dataportal.orr.gov.uk/statistics/performance/p-coded-cancellations/ · government ^[13] Bericht der Durchsetzungsstelle Fahrgastrechte des ... — https://www.era.europa.eu/sites/default/files/2025-09/t%C3%A4tigkeitsbericht%20eisenbahn%202023_2024_0.pdf · government ^[14] Passenger rail performance – October to December 2024 — https://dataportal.orr.gov.uk/media/p41ee2qs/performance_stats_release_2024-25_q3.pdf · government ^[16] Ergebnisse nach TU QMS TRV 2025: Résultats par ET SRQ ... — https://www.bav.admin.ch/dam/de/sd-web/Im-ByQHWXzdV/qms_rpv_2025_resultate_nach_tu.pdf · government ^[17] [PDF] Passenger rail performance – April to June 2024 - ORR Data Portal — https://dataportal.orr.gov.uk/media/ocib4lie/performance_stats_release_2024-25_q1.pdf · government ^[18] Railway Undertakings Service Quality Reports — https://www.era.europa.eu/library/railway-undertakings-service-quality-reports · government ^[20] Schweiz - SBB: «Die pünktlichste Bahn Europas» — https://www.srf.ch/news/schweiz/schweiz-sbb-die-puenktlichste-bahn-europas · professional ^[21] UK rail firms ordered to stop abusing train cancellations loophole — https://www.theguardian.com/uk-news/2023/jan/19/uk-rail-firms-ordered-stop-abusing-train-cancellations-loophole · professional ^[23] Deutsche Bahn mit Milliardenverlust und so unpünktlich wie noch nie — https://www.tagesschau.de/wirtschaft/unternehmen/deutsche-bahn-milliarden-verlust-unpuenktlich-100.html · professional ^[24] European Passengers' Federation – Punctuality report of European trains — https://www.epf.eu/wp/10929-2/ · professional ^[25] SBB: Noch nie so pünktlich wie 2024 — https://www.20min.ch/story/932-prozent-rekordwert-die-sbb-war-noch-nie-so-puenktlich-wie-2024-103268575 · professional ^[26] Deutsche Bahn: Wurde die Pünktlichkeitsstatistik manipuliert? — https://www.zdfheute.de/wirtschaft/unternehmen/deutsche-bahn-verspaetung-puenklichkeit-statistik-zugausfall-manipulation-100.html · professional ^[27] Bahn: Pünktlichkeit im Fernverkehr weiter sehr schlecht — https://www.br.de/nachrichten/deutschland-welt/bahn-puenktlichkeit-im-fernverkehr-weiter-sehr-schlecht,UsX03gU · professional ^[28] Pünktlichkeit der Bahn verschlechtert sich im Juli deutlich — https://www.tagesschau.de/wirtschaft/unternehmen/deutsche-bahn-unpuenktlichkeit-ziele-verfehlt-100.html · professional ^[29] Definitions of Railway Performance Metrics — https://www.raildeliverygroup.com/publications/acop/287-rdgntfdorpm-final/file.html · professional ^[30] Pünktlichkeit bei der Deutschen Bahn auf neuem Tiefpunkt — https://www.tagesschau.de/inland/deutsche-bahn-puenktlichkeit-102.html · professional ^[31] Rekordzahlen bei der SBB - SBB steigert im Jahr 2024 Pünktlichkeit weiter — https://www.srf.ch/news/schweiz/rekordzahlen-bei-der-sbb-sbb-steigert-im-jahr-2024-puenktlichkeit-weiter · professional ^[32] Pünktlichkeit im Bahn-Fernverkehr auf 60 Prozent gesunken — https://www.tagesschau.de/wirtschaft/deutsche-bahn-puenktlichkeit-104.html · professional ^[34] BAV-Bericht 2025: Pünktlichkeit und Qualität im regionalen Personenverkehr — https://www.mobilityblog.ch/bav-bericht-2025/ · general ^[35] Erläuterung Pünktlichkeitswerte für den August 2026 — https://www.deutschebahn.com/de/konzern/konzernprofil/zahlen_fakten/puenktlichkeitswerte-6878476 · general ^[36] Wie viele Bahnreisende pünktlich ans Ziel kommen - Matthias Gastel — https://www.matthias-gastel.de/wie-viele-bahnreisende-puenktlich-ans-ziel-kommen/ · general ^[37] DB Fernverkehr AG Geschäftsbericht 2024 — https://ir.deutschebahn.com/fileadmin/Bildmaterial/2024/Toechter_2024/DB24_Fernverkehr_web_01.pdf · general ^[38] Deutsche Bahn Annual Report 2024 — https://ibir.deutschebahn.com/2024/en/combined-management-report/product-quality-and-digitalization/the-customer-is-at-the-center-of-our-actions/punctuality/ · general ^[39] Deutsche Bahn Integrierter Bericht 2024 — https://ir.deutschebahn.com/fileadmin/Deutsch/2025/DB_IB24_d_web_01.pdf · general ^[40] Passenger Rail Performance 2020-21 Quarter 3 - PRgloo — https://cdn.prgloo.com/media/download/690d117b3c024130be9b2a0b3acb9b89 · general ^[41] Pünktlichkeit Deutsche Bahn 2024 | Deutsche Bahn Integrierter Bericht 2024 — https://ibir.deutschebahn.com/2024/de/zusammengefasster-lagebericht/produktqualitaet-und-digitalisierung/kunde-im-mittelpunkt-unseres-handelns/puenktlichkeit/ · general

Source quality: 1 academic, 14 government, 12 professional, 8 general.

Verification

• GB 84.8% Time to 3 in FY 2024-25, -0.4pp YoY, +0.8pp vs 2019-20, 92.9% in FYE March 2021 not in cards; only GB Time to 3 card is 85.0% in latest quarter 1 April to 30 June 2026.
• GB calendar-2024 cuts 67.0% On Time 54.8m/81.7m 12m to Dec 2024, 62.1% 12.7m/20.4m Q4 2024, PPM 85.2% 12m and 81.4% Q4, plus Apr-Jun 2024 70.1% and 87.4% not in cards.
• Germany 2024 rates 89.5% DB passenger, 89.4% DB Group, 88.1% all-rail, 90.7% DB Regio, 62.5% Fernverkehr operational down from 64.0%, 67.4% traveller, 4.0% cancelled, 10.2 min avg delay, 71.5% target missed not in cards.
• Switzerland 93.2% SBB trains +0.7pp record, Region Mitte 93.71%, Ost 93.65%, West 91.9% +2.7pp, Ticino 92.6% +2.1pp, connection 98.7%, freight 88.0% -1.8pp, 22 Nov 2024 snowfall not in cards; cards only have 13 of 23 above minimum and better availability of facilities and rolling stock contributed.
• Cancellations denominator: 4.1% FY 2024-25 record high, Q4 2024 5.1% official vs 5.7% with P*-coded, Apr-Jun 2024 3.5% vs 4.0%, 22:00 freeze, full=1 part=0.5 weighting, <50% vs >=50% mileage rules; cards confirm only weighting full as one part as half, 4.1% in latest quarter, and P-coded confirmable as late as 10pm previous evening.
• Thresholds and measuring points: GB On Time <1 min 59s, Time to 3 <3 min 2:59, PPM 5 min/10 min Long Distance, every recorded stop equal weight, early=on time, ~96% recorded with no estimate for ~4%; DB >800,000 runs/month >20,000 long-distance + ~780,000 local, >10m passengers/month; SBB <3 min arrival at destination/station plus connections maintained; cards do not contain these splits.
• Comparability warnings attributed to ORR limited direct comparisons ^[14][17] and EPF difficult definitions differ ^[24] have no verbatim card text.
• SBB regional bar chart with % values 93.2, 93.71, 93.65, 91.9, 92.6 sourced to ^[31][25] is not source-supported by provided cards and risks implying precision without primary SBB release in evidence.
• Anachronistic methodology: DB Erläuterung August 2026 ^[35] for <6 min/<15 min rules and BAV QMS RPV 2025 dated May 2026 ^[16] with 93.00 points minimum used to interpret 2024; draft discloses dating but still relies on them.
• Unresolved gaps acknowledged but still missing: no calendar-2024 GB Time to 3 annual average; Swiss cancellation inclusion, timetable-freeze rule, recording completeness, national all-operator share, 2024 customer punctuality; German whole-network cancellation rate, early-train handling, frozen-timetable rule.