How much do Uber drivers make per mile? And why we deleted that test from our own app

Dollars per mile is the number drivers reach for first, and it is easy to see why. Miles are what wears the car out, miles are what burns the fuel, and dividing the payout by the miles feels like it strips the marketing off the offer card.

It is also the reason a lot of good offers get passed on.

I know this because DriverSignal's decision engine used to include a per-mile test, and we took it out. Not tuned it, took it out. The reason is worth walking through, because it is the same reason the number misleads drivers who are doing the maths in their heads.

The number has no clock in it

Pay per mile answers the question "how much for this distance". It has no opinion at all about how long the distance will take.

Look at two offers with identical payouts and identical distances. Illustrative figures, but the shapes are ordinary.

Offer A: $16.50, 15.0 miles, 20 minutes, with a 1.0 mile pickup at 3 minutes. That is a run out along a highway at a steady speed.

Offer B: $16.50, 15.0 miles, 48 minutes, same 1.0 mile pickup. Same distance, but stop-start through a city centre at five in the afternoon.

On dollars per mile these two offers are indistinguishable. Both work out at $1.10 per trip mile. A driver filtering on that measure treats them as the same offer.

Now put the clock in. Total miles 16.0 in both cases, so at $0.30 a mile the driving cost is $4.80 and the net pay is $11.70, again in both cases.

| | Offer A | Offer B | |---|---|---| | Payout | $16.50 | $16.50 | | Pay per trip mile | $1.10 | $1.10 | | Net pay after costs | $11.70 | $11.70 | | Total time | 23 min | 51 min | | Net dollars per hour | $30.52 | $13.76 |

Same payout, same distance, same rate per mile, and one of them is worth more than twice the other. If you are holding a $20 an hour floor, one is an easy take and the other is not worth it.

There is no version of dollars per mile that can see that difference, because the difference is entirely made of minutes.

The failure that got the rule deleted

The mistake runs the other way too, and that is the one that broke our engine.

Long trips almost always price lower per mile than short ones. So a per-mile threshold, whatever you set it to, systematically flags long trips as bad.

Take an offer of $34.00 for 42 miles and 46 minutes, with a 2.0 mile pickup at 5 minutes. That is $0.81 per trip mile. Against a rule that wants a dollar a mile, it fails, and the app shows red.

Run it properly. Total miles 44.0, cost at $0.30 a mile is $13.20, net pay $20.80, total time 51 minutes. That is $24.47 an hour, comfortably over a $20 floor.

We were marking that offer as one to skip. A highway run that clears your floor, gets you a long way in under an hour, and puts almost no stop-start wear on the brakes, and the app was reddening it because a ratio said so.

That is not a threshold that needed adjusting. Any threshold high enough to catch genuine lowball offers is also high enough to condemn long trips, because the two look the same on that axis. So the test came out, and the engine now decides on net dollars per hour, which is what a driver is actually selling.

Dollars per mile can still be shown as one number among several. It is never allowed to be the gate.

Why the instinct is so hard to shake

The per-mile habit did not come from nowhere. Freight is priced per mile. Delivery contracts are priced per mile. The IRS reimburses per mile. If you came to rideshare from any of those worlds, per-mile thinking is the correct instinct carried over from a job where it worked.

It worked there because in freight the miles are the job and the clock is roughly proportional to them. A loaded trailer covers highway miles at a predictable speed, so per-mile and per-hour rank offers in nearly the same order.

Rideshare breaks that proportionality on purpose. A downtown pickup at 5pm and a suburban airport run are the same distance and completely different jobs. The moment the relationship between miles and minutes stops being stable, the per-mile shortcut stops agreeing with the thing it was standing in for, and you are left holding a ratio that no longer answers the question you asked it.

The two per-mile numbers people mix up

There are two dollars-per-mile figures in rideshare driving and confusing them causes a lot of bad arithmetic.

Cost per mile is what the car takes. DriverSignal defaults to $0.26 a mile, assembled from 28 MPG at $3.20 a gallon for fuel, roughly $0.11, plus around $0.15 for tyres, brakes, oil and depreciation.

That $3.20 is a default setting rather than a reading of the pump. EIA's published average for regular gasoline was $4.096 a gallon nationally for the week of 2026-07-27, which at 28 MPG is $0.146 a mile of fuel and a total nearer $0.30. The gap between cities is wider still: the same EIA week had Houston at $3.600 and San Francisco at $5.578.

Pay per mile is what the offer gives you for the same distance.

Subtract one from the other and you get margin per mile, which is a genuinely useful figure for a week or a month. It is still useless for judging a single offer, because margin per mile has no clock in it either. The full build of the cost side is in what a mile actually costs you, and the fuel half changes a lot with the car: an EPA-rated 52 MPG Prius sits at roughly half the fuel cost per mile of a 26 MPG Camry.

What the number is genuinely good for

I am not telling you to throw it away. Three jobs it does well.

Sizing your wear over time. Total miles for the month times your cost per mile is the amount your car quietly billed you. It is the number most drivers have never calculated and the one that explains why the bank balance never matches the effort.

Put figures on it. Say the odometer moved 2,800 miles across a month of driving. At $0.30 a mile that is $840 the car took out of the month before you paid yourself anything, and only a part of it left your pocket at a pump where you could feel it. The rest went into tyres you will buy in October and a trade-in value you will discover next year. Nothing in the Uber app will ever show you that $840. Multiply your own monthly mileage by your own cost per mile once and it stops being abstract.

Ranking two offers of similar length. If both cards are around 20 minutes, comparing pay per mile is a fast proxy and it will usually agree with the proper calculation. The proxy fails exactly when the durations differ, which unfortunately is most of the time.

Spotting a change in your market. If your pay per mile drifts down across a month while your hours hold steady, something moved. That is a real signal, and it is easier to see in the per-mile figure than in a weekly total.

None of those is a decision on the offer in front of you. That decision needs the clock.

What to use instead, in the seconds you get

Three lines, and they work on any offer from either app:

Total miles = pickup miles + trip miles Net pay = payout minus (total miles times your cost per mile) Net dollars per hour = net pay divided by (pickup minutes + trip minutes, in hours)

Compare the result against a floor you set in advance and hold. Mine started at $20 an hour, net, because it was a number I could keep in my head at a red light. After around fifty scored offers a floor learned from your own stream beats one you picked on day one, and there is a walkthrough in pick a floor and hold it.

The other thing worth carrying is a pickup limit, because deadhead is the input that per-mile thinking hides most effectively. Over 6.0 miles or 12 minutes of pickup and I pass, unless the offer still clears 1.5 times the floor after every cost. That reasoning is in the long pickup piece, and the hourly side of the argument is in how much drivers make per hour.

You can run any card through the calculator to check yourself. DriverSignal does the same arithmetic on the offer already on your screen, on Uber and Lyft, and shows the net and the net per hour before the card times out. It never touches Accept or Decline, so the call is yours every time, which is the only version of this I am willing to build.

Dollars per mile will still be sitting on that screen. Read it the way you read a tyre pressure: useful, worth knowing, and not what decides where you drive.