20 (Bellevue GE, Valavran – Veyrier, Tournettes)
Line 85:881:20 · Transports Publics Genevois · bus
Timetable period 2025
The published figure is the most recently completed timetable period. The current one is below it as a trend — month by month, and never folded into a single figure: nothing is averaged across a timetable change.
December 8, 2024 – December 13, 2025
74.89% — less than 3:00 late, coverage 99.19% (floor 90%)
| Threshold | Share | Coverage | Stop events counted |
|---|---|---|---|
| under 3:00 | 74.89% | 99.19% | 547'712 |
| under 5:00 | 86.51% | 99.19% | 547'712 |
| under 15:00 | 96.75% | 99.19% | 547'712 |
Distribution of delays
| Measure | Value (min:s) |
|---|---|
| Median | 1:13 |
| 90th percentile | 5:29 |
| 95th percentile | 7:56 |
| Measured arrivals | 536'296 |
The 3:00 threshold can see none of what happens beyond it: nine arrivals in ten are under 5:29.
Cancellations and early departures
| Measure | Share | Coverage |
|---|---|---|
| Ran rather than cancelled | 97.92% | 99.19% |
| Departures not early | 93.46% | 99.19% |
An early departure is one more than a minute ahead of the timetable. It has its own denominator — measured departures — because sharing the arrivals' one would make it depend on how many origin stops an operator has.
What this figure is made of
Three disclosures travel with every number on this page. For some operators they are not decoration — they are the whole finding.
| Disclosure | Share | What it means |
|---|---|---|
| Confirmed | 98.91% | The share of counted arrivals for which an arrival event fired (REAL). The rest is a forecast — which says something about a stop's equipment, not about whether the vehicle came. |
| Restated as timetable | 0.06% | The share of what could have been measured that was removed because the forecast was the scheduled second exactly. A denominator that lost a fifth of itself is a different statement from one that lost nothing. |
| Confirmed and exactly on plan | 0.5% | The share of confirmed arrivals filed at exactly the scheduled second. Disclosed and deliberately not removed: no rule in this project reaches this case. |
Only 3 of the 11 months in this period carry figures. The figure above is counted over those 3 and not over the whole period.
The months inside this period
The figure above is counted over the whole period. A single month inside it can have quite different coverage — and one that is below the floor on its own disappears into a period figure that clears it.
| Month | Share | Coverage | Confirmed | Restated as timetable | Confirmed and exactly on plan |
|---|---|---|---|---|---|
| September 2025 | 77.98% | 99.65% | 98.24% | 0.09% | 0.65% |
| October 2025 | 74.66% | 99.73% | 98.98% | 0.06% | 0.49% |
| November 2025 | 75.6% | 98.76% | 99.13% | 0.06% | 0.38% |
Rows with no share are below the 90% floor on their own. The figure above stands: it is measured over the whole period and it clears the floor. Both are true, and only one of them used to be on this page.
Nothing was reported under this line key in these months: January 2025, February 2025, March 2025, April 2025, May 2025, June 2025, July 2025, August 2025
Trend — Timetable period 2026
The current timetable period, month by month. Each month is counted on its own; these rows do not add up to the figure above and are not meant to.
| Month | Share | Coverage | Confirmed | Restated as timetable | Confirmed and exactly on plan |
|---|---|---|---|---|---|
| January 2026 | 78.5% | 99.21% | 99.3% | 0.03% | 0.38% |
| February 2026 | 76.31% | 99.37% | 99.64% | 0.02% | 0.33% |
| March 2026 | 73.21% | 91.63% | 98.96% | 0.08% | 1.64% |
| April 2026 | 66.69% | 97.21% | 99.3% | 0.03% | 0.51% |
| May 2026 | 62.29% | 95.99% | 97.86% | 0.14% | 0.39% |
| June 2026 | 61.56% | 96.18% | 98.73% | 0.08% | 0.45% |
| July 2026 | 70.33% | 99.46% | 99.48% | 0.04% | 0.32% |
| August 2026 | 72.58% | 99.23% | 99.05% | 0.07% | 0.51% |
| September 2026 | 61.93% | 98.72% | 99.13% | 0.07% | 0.27% |
Every row carries its own coverage. A month below the floor publishes no share — the three columns on the right still say why.
Not measured yet — these months are after the archive's most recent one: October 2026, November 2026
By hour of day
When an hour starts decides more about punctuality than almost anything else measured here.
Over the last 365 days in the archive, not the timetable year above — so the grid shows the hours as they run now.
October 1, 2025 – September 30, 2026
| Hour | Mon | Tue | Wed | Thu | Fri | Sat | Sun | Stop events counted |
|---|---|---|---|---|---|---|---|---|
| 04:00 | too few measurements | too few measurements | too few measurements | too few measurements | too few measurements | too few measurements | too few measurements | 487 |
| 05:00 | 87.29% | 82.93% | 87.29% | 86.43% | 88.74% | 91.34% | 86.48% | 64'196 |
| 06:00 | 88.64% | 88.29% | 91.44% | 89.22% | 90.43% | 90.6% | 86.86% | 84'709 |
| 07:00 | 74.72% | 74.28% | 80.27% | 77.11% | 82.15% | 93.44% | 89.58% | 120'752 |
| 08:00 | 57.25% | 53.6% | 68.32% | 57.19% | 73.7% | 88.74% | 86.66% | 131'882 |
| 09:00 | 60.04% | 52.42% | 66.17% | 55.94% | 73.46% | 80.75% | 82.23% | 120'775 |
| 10:00 | 70.8% | 63.93% | 63.47% | 64.99% | 72.52% | 82.45% | 78.44% | 100'026 |
| 11:00 | 76.32% | 70.54% | 66.64% | 72.11% | 70.74% | 77.9% | 84.95% | 98'756 |
| 12:00 | 77.2% | 72.31% | 60.97% | 69.96% | 70.67% | 79.71% | 81.99% | 100'987 |
| 13:00 | 74.36% | 71.35% | 64.63% | 68.82% | 69.83% | 83.1% | 80.88% | 102'913 |
| 14:00 | 70.46% | 67.67% | 62.74% | 67.58% | 67.24% | 80.94% | 82.32% | 104'725 |
| 15:00 | 73.84% | 68.26% | 65.85% | 68.08% | 66.53% | 81.03% | 79.89% | 103'928 |
| 16:00 | 63.63% | 59.11% | 65.86% | 59.57% | 60.16% | 78.4% | 81.57% | 127'864 |
| 17:00 | 55.3% | 47.64% | 61.56% | 50.8% | 62.14% | 80.23% | 79.08% | 135'892 |
| 18:00 | 52.31% | 39.53% | 55.3% | 44.81% | 63.53% | 76.78% | 76.78% | 139'391 |
| 19:00 | 62.84% | 48.28% | 64.02% | 48.3% | 62.68% | 80.89% | 79.52% | 119'751 |
| 20:00 | 74.49% | 72.66% | 79.15% | 66.92% | 69.97% | 82.4% | 78.95% | 89'928 |
| 21:00 | 86.19% | 81.54% | 78.18% | 73.18% | 76.93% | 80.7% | 79.92% | 69'813 |
| 22:00 | 89.24% | 82.29% | 78.79% | 75.78% | 73.51% | 76.73% | 80.64% | 58'686 |
| 23:00 | 88.37% | 81.24% | 78.69% | 74.9% | 64.3% | 72.03% | 80.24% | 60'368 |
| 00:00 (+1) | 89.82% | 88.17% | 85.55% | 84.57% | 71.29% | 77.29% | 88.35% | 55'834 |
| 01:00 (+1) | 87.81% | 88.58% | 89.01% | 84.13% | 77.49% | 77.35% | 76.64% | 15'617 |
| 02:00 (+1) | nothing measured | nothing measured | too few measurements | too few measurements | 91.79% | 86.49% | nothing measured | 4'655 |
| 03:00 (+1) | nothing measured | nothing measured | too few measurements | too few measurements | too few measurements | too few measurements | nothing measured | 1'080 |
| 04:00 (+1) | nothing measured | nothing measured | nothing measured | nothing measured | nothing measured | too few measurements | nothing measured | 16 |
The outlined cell is that day’s worst hour — it is not the same hour on every day.
Service day, not calendar day: hour 24 is 00:00–00:59 the following morning and still belongs to the day before.
An hour's coverage is the measurement axis alone. The figure at the top of this page takes the worst of three axes — days, journeys, measurements — and so rests on more evidence than these rows do.
About this line
| Line key | 85:881:20 |
|---|---|
| Label | 20 |
| Operator | Transports Publics Genevois |
| Mode | bus |
| Earlier keys | 85:881:TPG020 |
A line key changes when an operator renumbers its lines or the timetable issues new identifiers. Only a succession the registry has accepted is linked here — a proposed one is not.
The line registry is built from one week of June 2026. Lines that did not run in that week have no name here — they are measured all the same.
Source and method
Source of the measurements: opentransportdata.swiss (istdaten). How we measure is written up in the methodology.
A figure is published only from 90% coverage upward. Below it there is no figure here — not a greyed-out one, not a dash, none.