George Russell was the most consistent driver in Formula 1 in 2025. His lap times varied by 0.220 seconds. Andrea Kimi Antonelli, in the identical car, varied by 0.453.
That is a factor of 2.06, and it is the largest team-mate gap on the grid.
Across the whole field the range runs from 0.220 to 0.691 seconds, and the gap between the six-driver rookie class and everyone else turns out to be far smaller than you would expect.
Standard deviation of lap time residuals by driver across the 2025 season
Why the obvious method does not work
The usual way to measure consistency is to take a driver's race lap times and calculate the standard deviation. It is simple, and it mostly measures things that have nothing to do with the driver.
A Formula 1 car gets faster all race as it burns fuel, at roughly 0.3 to 0.4 seconds a lap for every 10kg. It gets slower as its tyres wear. It loses about half a second a lap whenever it is running within a second of another car, which we measured separately. Raw lap-time scatter is dominated by all three.
So a driver who happens to run in clean air on a one-stop strategy looks "consistent", and a driver who spent the afternoon in traffic looks erratic, regardless of how either actually drove.
We stripped all three out.
For every driver-stint we fitted a straight line through the lap times. That line absorbs fuel burn-off, which makes the car faster, and tyre degradation, which makes it slower. The residual is how far each lap fell from that driver's own expected pace at that moment.
Then we kept only clear-air laps, where the car ahead was more than four seconds away, so traffic cannot contaminate it.
The standard deviation of what is left is as close to driver consistency as lap times can get.
A useful check on all of this: published figures for F1 lap time variation put the range at roughly 0.2 to 0.7 seconds. Our numbers run from 0.220 to 0.691. We arrived there by a different route and landed on the same scale.
The 2025 order
Russell — 0.220s
Verstappen — 0.297s
Piastri — 0.329s
Bearman — 0.361s
Leclerc — 0.394s
Alonso — 0.401s
Gasly — 0.408s
Sainz — 0.448s
Stroll — 0.450s
Antonelli — 0.453s
Ocon — 0.457s
Hamilton — 0.465s
Lawson — 0.467s
Norris — 0.468s
Albon — 0.496s
Tsunoda — 0.502s
Bortoleto — 0.512s
Hulkenberg — 0.525s
Hadjar — 0.548s
Colapinto — 0.691s
Russell's margin at the top is the striking part. The gap between first and second, Russell to Verstappen, is 0.077 seconds. The gap between second and sixth, Verstappen to Alonso, is 0.104. One driver is separated from the field by roughly as much as the next five are separated from each other.
Same car, and the gap is still there
Consistency is affected by the car. A machine that is difficult on its tyres, or nervous under braking, will make anyone look untidy. The only way to hold that constant is to compare drivers who share it.
Consistency compared between team-mates across the 2025 season
Piastri was 1.42 times more consistent than Norris, in the car that won the constructors' championship, during a season in which the two of them were team-mates fighting at the front. Whatever else separated them, one of them put the car in the same place lap after lap more reliably than the other.
Kick Sauber is a dead heat. Bortoleto, a rookie, and Hulkenberg, one of the most experienced drivers ever to race in Formula 1, are separated by 0.013 seconds. That is nothing.
The rookie penalty is smaller than you would expect
2025 had an unusually large rookie class: Antonelli, Bearman, Bortoleto, Doohan, Hadjar and Lawson, six full-season debutants, roughly 30% of the grid.
The expectation is obvious. Rookies should be scrappier, less able to hit the same marks lap after lap.
The 2025 rookie class compared with established drivers on lap time consistency
Rookies averaged 0.468 seconds. Established drivers averaged 0.437. The gap is 0.031 seconds.
That is a real difference and a small one, about 7%. And it is not uniform:
Bearman was the fourth most consistent driver on the entire grid, ahead of Leclerc, Alonso, Hamilton and Norris, in his first full season.
Bortoleto matched Hulkenberg in the same car.
Hadjar at 0.548 and Antonelli at 0.453 sit in the bottom half, which is closer to the expectation.
The lesson is not that experience does not matter. It is that the consistency gap between a good rookie and an established driver is much smaller than the gap between the best and worst drivers overall. Russell to Bearman is 0.141 seconds. Bearman to Colapinto is 0.330.
What this measures, and what it does not
Consistent is not the same as fast. This measures how tightly a driver clusters around their own pace, not what that pace is. A driver could lap consistently slowly and score well here. Read it alongside our team race pace measurements, not instead of them.
Some of it is still the car. The per-stint fit removes the average degradation trend, but a car whose tyres fall away suddenly rather than smoothly will produce larger residuals for whoever is driving it. The team-mate comparison is the defence against this, and it is why that chart matters more than the outright ranking.
Clear-air laps are not evenly distributed. Drivers at the front get more of them. Russell has 352 clear-air laps in this sample and Verstappen 218. More laps means more chances to record an outlier, so Russell's figure is if anything harder to achieve, not easier.
Fuel saving and tyre management are real driving. A driver instructed to lift and coast will show larger residuals. That is a team decision showing up as driver inconsistency, and lap times cannot separate the two.
Method
Data: FastF1, 2025 race sessions.
For each driver-stint, a linear fit of lap time against lap number, which absorbs fuel burn-off and tyre degradation together. Consistency is the standard deviation of the residuals against that fit.
Clear-air laps only, defined as the car ahead being more than 4.0 seconds away, measured from cumulative end-of-lap times.
Excluded: in-laps, out-laps, laps 1 and 2, any lap not under green flag, deleted laps, and residuals beyond 5 seconds.
Drivers need at least 120 qualifying clear-air laps to appear. That admits 20 drivers; Jack Doohan did not race enough of the season to qualify, so five of the six rookies are represented.
Sample sizes vary a lot, from 160 laps (Albon) to 443 (Leclerc). Figures from smaller samples are less precise.
The 0.2 to 0.7 second range quoted as a cross-check comes from published analyses, not from us, and is cited as an external sanity check on scale rather than as a source for any figure here.
