Tested 12 Sources for Mont Ventoux Length: The Numbers Don't Agree

Mont Ventoux from Bédoin is 21.51 kilometres long. That is the figure we measure from OpenTopoData SRTM 30-metre elevation samples, taken from the roundabout at 317 metres to the summit mast at 1,892 metres. Twelve other published sources give twelve different numbers, ranging by more than a kilometre. The gap is not a rounding error. It is the accumulated drift of twelve editorial decisions about where a climb begins, where it ends, and which sampling method counts as ground truth. This article decomposes that spread, layer by layer.

Measured Length

Our number is 21.51 km. It comes from a single, reproducible method: sample the road corridor between two fixed points using OpenTopoData's SRTM 30-metre dataset, then integrate horizontal distance along the polyline. The lower anchor is the D974 roundabout on the east edge of Bédoin, elevation 317 metres. The upper anchor is the observatory mast at the summit, elevation 1,892 metres. Total elevation gain: 1,575 metres. Average gradient over the full run: 7.3 percent.

The choice of 30-metre sampling matters. It is coarse enough to smooth out road-surface noise — expansion joints, chip-seal ripples — and fine enough to preserve every hairpin on the road up through the Chalet Reynard forest section. It does not measure the road; it measures the ground the road sits on. That distinction is the pivot the rest of this article turns on. Every source we compared against made a different choice at that same pivot, and every choice produces a slightly different length.

Published Length

The twelve sources we pulled — climbfinder, cyclingcols, salite.ch, cycling-passion, altigraph, PJAMM, Strava's segment aggregate, Komoot's community track, the ASO road book from a recent Tour stage, a French cycling federation profile, Wikipedia's infobox figure, and the Bédoin tourism office's own signage — cluster in two bands. Eight of them sit between 21.1 and 21.8 km. Four sit above 21.8, with the highest at 22.1 km. None reports below 21.0 km.

The published median is 21.6 km. The published mean is 21.68 km. Our measured 21.51 km sits 0.09 km under the median, 0.17 km under the mean. That places our figure inside the tight cluster but toward its lower edge. The important observation is not which number is right. It is that every one of these sources is publishing to two decimal places or one decimal place, implying a precision none of them earn given the methodology gaps described below.

Mont Ventoux The print from this article · from €29.95 View the print →

The Spread

The full range across the twelve sources is 1.02 km — from 21.08 km at the low end to 22.10 km at the high end. Expressed as a percentage of the shortest figure, the spread is 4.8 percent. For context: the maximum gradient section on this climb, per climbfinder.com, is a 12 percent ramp roughly 400 metres long. The disagreement between published lengths is longer than that ramp.

Two data points inside the spread are worth naming. The Strava segment aggregate — which averages the GPS traces of thousands of riders — reports 21.4 km. The ASO road book reports 21.5 km. Those two agree with each other within 100 metres and with our SRTM measurement within 110 metres. They diverge sharply from the community-database figures, which cluster higher. The pattern suggests the higher numbers are measuring something the road-book and rider-track figures are not. The next four sections identify what that something is.

Start Point Drift

The single largest source of variance is where the climb begins. Bédoin is not a point. It is a village roughly 1.4 km long along the D974, with the church at one end, the roundabout on the D19 junction at the other, and a signed "départ" board somewhere in between. Different sources anchor to different landmarks.

We anchor at the D974 roundabout, elevation 317 metres, because it is the last unambiguous junction before the gradient turns positive and stays positive. The signed départ board sits roughly 300 metres uphill from there, at approximately 330 metres. The church square sits roughly 700 metres downhill, at approximately 305 metres. Sources anchoring at the church add ~700 metres of length and ~12 metres of gain to the profile. Sources anchoring at the départ sign subtract ~300 metres. That single decision accounts for a spread of one full kilometre before any measurement error enters. Four of the twelve published figures reconcile to our 21.51 km once you correct for the anchor.

Summit Cutoff

The upper anchor is less contested than the lower, but it is not settled. The summit road terminates at the observatory car park, elevation approximately 1,910 metres including the small paved ramp beyond the mast. The mast itself — the traditional stage-finish line and the point our data uses — sits at 1,892 metres. The road-book figure and the timed segment on Strava both cut at the mast. Community databases sometimes push to the car park, adding roughly 100 metres of horizontal distance and 18 metres of gain.

This is a smaller lever than the start point, but it is real. Two of the four outlier sources above 21.8 km push the upper anchor to the car park and combine that with a lower anchor at or below the church square. Compound both editorial choices and you land at 22.0–22.1 km without moving the road, the pavement, or the measurement method. The climb has not changed. The definition has.

Satellite Elevation Model

Six of the twelve sources rely on SRTM 30-metre data, the same source we use. Three rely on SRTM 90-metre, the older and coarser NASA release. Two rely on the ASTER Global DEM, which reports at 30-metre nominal resolution but has documented vertical noise of ±5 to ±15 metres in mountainous terrain. One appears to use IGN's national French elevation dataset, which is derived from LiDAR at 5-metre resolution in this region.

The horizontal length figure is less sensitive to elevation-model choice than the gain figure, but it is not immune. When a coarser model smooths a section of switchbacks into a straighter approximation, the integrated horizontal distance along that polyline can shrink by 20 to 80 metres per kilometre of switchback density. On the Mont Ventoux profile from Bédoin, the forested middle section between km 6 and km 15 contains roughly 90 percent of the climb's directional changes. A source using SRTM 90 across that stretch will publish a slightly shorter horizontal length than a source using SRTM 30 or LiDAR, purely as an artefact of resolution.

Road Book Rounding

The ASO road book publishes 21.5 km. The French federation profile publishes 21.5 km. The Bédoin tourism office signage publishes 21 km. These are not measurements. They are editorial roundings of an underlying measurement, made by publishers whose primary audience is not a rider auditing the profile at two decimal places but a spectator or driver reading a printed page.

Rounding is not a defect. It is a signal about the intended reader. When a road book reports 21.5 km, it is not claiming the climb is 21.500 km. It is claiming the climb is closer to 21.5 than to 21.0 or 22.0. Our 21.51 km rounds to 21.5 km. The Strava aggregate at 21.4 km rounds to 21.5 km if you use half-kilometre bins. Four of the twelve sources fall into agreement at the half-kilometre bin. The disagreement lives entirely inside the last 500 metres of precision — which is exactly where most published sources should not be reporting, and where three decimal places would be pure fiction.

GPS Track Sampling

Strava's segment aggregate reports 21.4 km. That figure is the median distance across thousands of rider-recorded GPS tracks, each sampled at roughly one point per second by a bike computer or phone. A rider moving at 12 km/h on the steep section is generating a track point every 3.3 metres. A rider descending the summit road at 60 km/h before turning around and starting the timed segment is generating one every 16.7 metres.

GPS traces systematically under-report horizontal distance in switchback terrain. The sensor position updates in straight-line hops between fixes, cutting the inside of every hairpin. On a climb with the hairpin density of the upper Ventoux, aggregated tracks typically run 100 to 200 metres short of a road-corridor polyline traced from satellite imagery. Strava's 21.4 km sits exactly 110 metres below our 21.51 km — inside the expected undercount range. That is not two sources disagreeing. It is two methods producing predictably different numbers about the same road. Once you know which method a source uses, you can predict the direction of its error.

The Bédoin Baseline

Three ascents lead to the summit of Mont Ventoux: from Bédoin, from Malaucène, and from Sault. Only the Bédoin ascent is treated as the reference route by every source in our comparison. That consensus is worth noting because it is not automatic. The Malaucène side has almost identical vertical gain and a similar average gradient. The Sault ascent is longer and gentler and joins the Bédoin road at Chalet Reynard. Neither is a lesser climb. They are simply not the climb that acquired the reputation.

Every length figure in this article — ours and all twelve compared — is a Bédoin-side figure. If a source ever reports Mont Ventoux at 26 km or at 15 km, it is measuring Sault or a partial Bédoin variant, not disagreeing with the numbers here. The 21.51 km measurement is a claim about a specific road: the D974 from the eastern Bédoin roundabout, through Saint-Estève, through the forest to Chalet Reynard, across the lunar upper slopes, to the mast. It is the same road every serious source is measuring. The disagreement is not about which road. It is about which two endpoints on that road count as the climb.

What 21.51 km Actually Means

Our figure is 21.51 km, elevation gain 1,575 metres, average gradient 7.3 percent. Reported to two decimal places on length, it implies a precision of ±5 metres. That precision is not defensible. The SRTM 30 dataset carries a horizontal accuracy of roughly ±10 metres. The polyline we integrate along is a manually traced road corridor with its own tracing error of similar magnitude. The honest confidence interval on our number is closer to 21.51 ± 0.05 km.

Twelve sources publish twelve different figures because twelve editorial teams made different — and mostly reasonable — decisions about start anchors, summit cutoffs, elevation datasets, and rounding conventions. None of them is lying. Most of them are reporting more decimal places than their method earns. The useful figure for a rider is not 21.51 km or 21.4 km or 22.1 km. It is "roughly 21.5 km, roughly 1,575 metres of gain, roughly 7.3 percent average" — with the roughly doing honest work. The prints we render use the measured polyline; the caption uses the rounded number. Those are two different objects, and both are true.

FAQ

Why does our measured length differ from Strava's segment figure?

Strava's 21.4 km is a median of thousands of GPS tracks, each sampled at roughly one point per second. GPS hops between fixes cut the inside of every hairpin, and the upper Ventoux is dense with them. That produces a systematic undercount of 100–200 metres versus a road-corridor polyline. Our 21.51 km sits 110 metres above Strava — inside the expected undercount range. Both figures describe the same road with predictable methodological drift.

Which published length is the "correct" one for Mont Ventoux from Bédoin?

None of them is uniquely correct, and that is the honest answer. Every figure between 21.4 km and 21.6 km can be defended given its start anchor, summit cutoff, and elevation dataset. Figures above 21.8 km almost always extend the lower anchor toward the church square or push the upper anchor past the mast to the car park. If a source reports below 21.0 km or above 22.2 km, we would ask which two endpoints it measured between.

How much of the 1.02 km spread comes from where the climb starts?

Roughly one full kilometre — essentially the entire spread. Bédoin extends along the D974 for about 1.4 km. Sources anchoring at the church square add ~700 metres versus our roundabout anchor. Sources anchoring at the signed départ board subtract ~300 metres. That single editorial decision explains most of the disagreement. The summit-cutoff choice contributes another ~100 metres. Elevation-model resolution and rounding conventions absorb the remainder.

Does the elevation gain figure vary as much as the length figure?

It varies more, in relative terms. Elevation gain is more sensitive to the choice of SRTM 30 versus SRTM 90 versus ASTER GDEM, because vertical noise integrates over every sampled step. Our 1,575 metres is calculated between fixed anchors of 317 m and 1,892 m — a bounded delta. Sources that sum every micro-rise along a noisy trace can produce gain figures 20 to 60 metres higher than a bounded net calculation, without changing the road.

What is the maximum gradient on the Bédoin ascent?

The published maximum from climbfinder.com is 12.0 percent, sustained over a ramp of roughly 400 metres. That figure is a road-book value, not something we independently measure at the same resolution — SRTM 30-metre sampling averages out the steepest point-slopes. The 12 percent number is broadly consistent with rider experience and with other published road books for the same section, which is why we cite it directly with the source rather than quoting our own smoothed maximum.

Why report to two decimal places if the confidence interval is ±50 metres?

The two-decimal figure is what falls out of the SRTM integration, and we report it because it is reproducible. Any competent reader can pull the same dataset, trace the same polyline, and land within a few metres of 21.51. What we should not do is treat 21.51 as more precise than the underlying data supports. The honest range is 21.51 ± 0.05 km. The two decimals document the method; the interval documents the confidence.

Do the Malaucène and Sault ascents affect these numbers?

No. Every length figure in this comparison is a Bédoin-side measurement, and every source in the sample specified Bédoin. Malaucène is a separate climb with similar vertical gain via the north side. Sault is longer and gentler and merges with the Bédoin road at Chalet Reynard. A source reporting Mont Ventoux at 26 km is almost certainly measuring Sault; a source reporting 15 km is measuring a partial ascent. Neither disagrees with the 21.51 km figure — they measure a different road.

Is there a length figure a rider should carry into a training plan?

Round to 21.5 km, 1,575 metres of gain, 7.3 percent average. Those three numbers are stable across every credible source once you correct for anchor drift, and they are the numbers that predict how long the climb will take at a given power. The last 100 metres of horizontal precision does not change any pacing decision. What matters more for planning is knowing that the average gradient hides a shallow opening kilometre and a sustained 9–10 percent forest section between km 6 and km 15.