The archive
Thirteen years of L’Étape du Tour
10 editions, 107,795 classified finishers, and three years with no race at all. Every figure on this page is read from the timing provider’s own results and re-checked against the database that stores them.
Classified finishers per edition, from zero — so a point's height is the size of the field. The gaps are real, and not all the same: 2018 was raced but its results are absent from the timing provider's database, while 2020 and 2021 were cancelled outright.
The size of the field, 2014-2026. Use the left and right arrow keys to read each point.
2026 is the only edition whose feed publishes riders who did not finish, so it is the only year where an abandon rate could be computed at all.
Show the numbers
| Year | Finishers |
|---|---|
| 2014 | 8,459 |
| 2015 | 9,774 |
| 2016 | 10,982 |
| 2017 | 11,234 |
| 2018 | — |
| 2019 | 10,159 |
| 2020 | — |
| 2021 | — |
| 2022 | 8,706 |
| 2023 | 11,485 |
| 2024 | 10,614 |
| 2025 | 13,649 |
| 2026 | 12,733 |
You cannot compare the times. You can compare the speeds.
The Étape reproduces a Tour de France stage, so the course changes completely every year — 122 km at Morzine in 2016, 181 km to Col d'Izoard in 2017. A chart of finishing times across editions would be a chart of the parcours. So everything below is a rate.
Even then the ground keeps moving. The fastest winning average in the archive is 34.43 km/h (2017, Briançon → Col d'Izoard); the slowest is 27.91 km/h (2015). That gap says more about how much climbing was in the route than about how the riders have changed.
From 122 km in 2016 to 181 km in 2017. Comparing finishing times across editions would compare the parcours, so everything here is km/h — and the axis runs from zero, so the median rider's speed is genuinely two thirds of the winner's.
Speed, not time — because the course is never the same twice. Use the left and right arrow keys to read each point.
Show the numbers
| Year | Winner | Median rider | Slowest tenth |
|---|---|---|---|
| 2014 | 30.8 km/h | 17.0 km/h | 13.9 km/h |
| 2015 | 27.9 km/h | 14.0 km/h | 11.7 km/h |
| 2016 | 34.3 km/h | 17.7 km/h | 13.7 km/h |
| 2017 | 34.4 km/h | 18.7 km/h | 15.2 km/h |
| 2018 | — | — | — |
| 2019 | 28.5 km/h | 16.0 km/h | 13.0 km/h |
| 2020 | — | — | — |
| 2021 | — | — | — |
| 2022 | 31.5 km/h | 17.2 km/h | 14.5 km/h |
| 2023 | 32.0 km/h | 17.2 km/h | 13.9 km/h |
| 2024 | 29.5 km/h | 16.1 km/h | 13.2 km/h |
| 2025 | 29.6 km/h | 15.9 km/h | 12.8 km/h |
| 2026 | 31.8 km/h | 17.1 km/h | 14.3 km/h |
Three years with no race
The archive has holes, and they are not all the same kind of hole. Two are history; one is a gap in the data itself, which is a different claim and is worth keeping separate.
Raced — Annecy to Le Grand-Bornand, 169 km — but the timing provider holds no results for it. The event and the race both exist in the API, and the race returns zero rows; resolving the event by its internal id reaches the same empty race. The data is absent at source, not missing from this import.
Scheduled for 5 July at Nice, postponed to 6 September, then cancelled outright on 28 July 2020 — the first cancellation in the event’s history. Entrants received an 18-month credit. A separate virtual edition ran on Zwift alongside the Virtual Tour de France.
The deferred Nice edition was set for 4 July 2021 and cancelled in late May, in agreement with the City of Nice and following the Alpes-Maritimes prefecture. Entry fees were refunded. ASO ran its autumn road races that year, but L’Étape sat before that reopening and was not moved.
Edwige Pitel, first woman home four editions running
The same name appears at the top of the women’s results in 2015, 2016, 2017, 2019 — four consecutive editions of the race. Her best is 2017, when she came 55th of 11,234 on the 181 km to the Col d’Izoard, the longest course in this archive and the best placing by any woman in it. She was 48.
Share of classified finishers. This is the main Étape only — the Étape du Tour de France Femmes is a separate event with its own results.
Women in the field. Use the left and right arrow keys to read each point.
Best placing by a woman, by edition: 2014 Magdalena de Saint Jean (154th), 2015 Edwige Pitel (85th), 2016 Edwige Pitel (70th), 2017 Edwige Pitel (55th)…
Show the numbers
| Year | Women |
|---|---|
| 2014 | 3.3% |
| 2015 | 4.1% |
| 2016 | 5.4% |
| 2017 | 5.3% |
| 2018 | — |
| 2019 | 6.4% |
| 2020 | — |
| 2021 | — |
| 2022 | 4.4% |
| 2023 | 6.5% |
| 2024 | 6.5% |
| 2025 | 6.8% |
| 2026 | 5.2% |
Share of finishers registered with a French nationality. The line stops at 2025 on purpose.
How French the field is. Use the left and right arrow keys to read each point.
In 2026 the timing feed carries a nationality for only 4.5% of finishers. Plotting that year would show a cliff that is a missing column, not a change in who turned up — so it is left out rather than drawn.
Show the numbers
| Year | French |
|---|---|
| 2014 | 61.6% |
| 2015 | 64.1% |
| 2016 | 60.9% |
| 2017 | 65.3% |
| 2018 | — |
| 2019 | 69.7% |
| 2020 | — |
| 2021 | — |
| 2022 | 77.2% |
| 2023 | 79.1% |
| 2024 | 73.9% |
| 2025 | 77.5% |
| 2026 | — |
Every edition
| Year | Course | km | Finishers | Winner | Winning time | Median |
|---|---|---|---|---|---|---|
| 2014 | Pau → Hautacam Côte de Bénéjacq · Côte de Loucrup · Col du Tourmalet · Hautacam | 148 | 8,459 | Loïc Herbreteau | 4h 48m 25s | 8h 42 |
| 2015 | Saint-Jean-de-Maurienne → La Toussuire Col du Chaussy · Col du Glandon · Col de la Croix de Fer · Col du Mollard · La Toussuire | 138 | 9,774 | Jérémy Bescond | 4h 56m 38s | 9h 50 |
| 2016 | Megève → Morzine Col des Aravis · Col de la Colombière · Col de Joux Plane | 122(146.5 official) | 10,982 | Tao Quéméré | 3h 33m 35s | 6h 53 |
| 2017 | Briançon → Col d'Izoard Côte des Demoiselles Coiffées · Col de Vars · Col d'Izoard | 181 | 11,234 | Jonas Abrahamsen | 5h 15m 26s | 9h 40 |
| 2019 | Albertville → Val Thorens Cormet de Roselend · Côte de Longefoy · Val Thorens | 135 | 10,159 | Adrien Guillonnet | 4h 44m 11s | 8h 27 |
| 2022 | Briançon → L'Alpe d'Huez Col du Galibier · Col de la Croix de Fer · Alpe d'Huez | 166.8 | 8,706 | Stefan Kirchmair | 5h 17m 25s | 9h 41 |
| 2023 | Annemasse → Col de Joux Plane Col de Saxel · Col de Cou · Col du Feu · Col de la Jambaz · Col de la Ramaz · Col de Joux Plane | 145(157 official) | 11,485 | Artus Jaladeau | 4h 31m 28s | 8h 26 |
| 2024 | Nice → Col de la Couillole Col de Braus · Col de Turini · Col de la Colmiane · Col de la Couillole | 138 | 10,614 | Damien Jeanjean | 4h 40m 28s | 8h 33 |
| 2025 | Albertville → La Plagne Côte d'Héry-sur-Ugine · Col des Saisies · Col du Pré · Cormet de Roselend · La Plagne | 131 | 13,649 | Victor Vidal | 4h 25m 17s | 8h 13 |
| 2026 | Le Bourg d'Oisans → Alpe d'Huez Col de la Croix de Fer · Col du Télégraphe · Col du Galibier · Col de Sarenne | 170 | 12,733 | Luca Cavallo | 5h 20m 20s | 9h 55 |
What this archive will not tell you
How many people abandoned. Only the 2026 feed publishes riders who did not finish — 17,439 entrants against 12,733 classified. Every other edition lists finishers only, so its abandon rate would compute as zero. That is a property of the feed, not of the race.
Where the nationalities went after 2025. The 2026 results carry a nationality for 4.5% of finishers, against 100% in every earlier year. The French-share line stops where the data does.
Anything about 2014’s age categories. 99% of that edition’s rows have no category at all, and the handful that do use an incomplete vocabulary. It is left out rather than extrapolated from a neighbouring year.
Where the timing breaks down
Not every edition supports every page, and the reasons are worth stating rather than leaving a reader to infer them from a missing table.
The timing provider publishes no intermediate mats for any of these editions — only the finish. So they have no checkpoint pages and no climb leaderboards, unlike 2025 and 2026, and the archive compares them on finishing speed alone. The splits exist upstream for at least 2023 and 2024, but placing them needs a course profile per year that the feed does not carry.
Raced — Annecy to Le Grand-Bornand, 169 km — and the timing provider holds zero results for it. The event resolves through its internal id and returns an empty set. That is a gap in their database, not in this import.
Nationality is recorded for 4.5% of finishers where every edition from 2014 to 2025 has it for all of them. The French-share series therefore stops at 2025: carrying it into 2026 would draw a collapse in foreign entries that is an artefact of a missing column.
How the results were read
One detail decided whether half this archive was right or wrong. The timing API reports a gun time and a chip time for every rider, and which one the official classification uses changed: the editions from 2014 to 2017 are ranked on gun time, and everything from 2019 onwards on chip time. Since every edition started in waves — up to two hours and fifty minutes separate the first and last rider off the line in 2015 — reading the wrong clock would have sorted a whole field by its start pen.
So the basis was not taken from the API’s own flag, which is unreliable on the newer races. For each edition both candidate times were sorted and compared against the provider’s published ranking; the one that reproduced it was the one used. On the older editions gun time reproduces the classification for 100% of riders and chip time for under half a percent, and after 2019 that reverses.
About this project and its data. The results are Sport Innovation / TimeTo’s — the official timekeeper of the event. They were fetched once after the event and republished here with segment times, recomputed leaderboards and a page per athlete that the official results do not offer. The timing is their work; the analysis is mine, and any mistake in it is mine too. Course details come from Amaury Sport Organisation (A.S.O.), the organiser. Rankings are the provider's own: gun time through 2017, chip time from 2019, established against the published results for each edition. Athletes who asked to be anonymous in the official results stay anonymous here.
107,795 finishers across 10 editions · the 2026 explorer has splits, segments and pacing for the most recent edition.