MileGrade

Keeping & Replacing

Repair or Replace Calculator

You have an estimate in your hand and someone has already told you not to spend more than the car is worth. That is the wrong test. The question is what the repair costs per mile it buys you — and whether the other way of having a working car costs less over the same stretch.

The repair, and what it buys

The written estimate, parts and labour, including tax.

The honest guess, and the one the answer turns on. Everything below is arithmetic; this is a forecast.

What someone would hand you for it today, broken. This is cash you only get if you actually replace it — so it counts on that side alone.

Keeping this car — what the next stretch costs

What it actually returns, not the sticker — the MPG calculator works it out from a fill-up.

An older car usually costs less to insure. Six-month policy? Double it.

What ELSE it will need over this window — beyond the repair above. On a car already in the shop, this is rarely nothing.

After the months above. Enter 0 if you'll drive it until it's scrap — for an old car that is often the honest answer.

The replacement — optional, over the same window

Leave this empty and you still get what the repair costs per mile. Fill it in and the two paths are priced against each other over the same months and the same miles.

Price, sales tax and fees — everything you'd sign for, before financing.

Interest only, over the months above — not the payment. Blank if you'd pay cash. The auto loan calculator prints it.

What it actually returns, not the sticker — the MPG calculator works it out from a fill-up.

A newer car normally costs more to insure, and a lender will require full coverage.

Oil, tires, brakes. Lower on a newer car, but not zero — and warranty work is not maintenance.

After the same months. This is the biggest number on this side and the least knowable — a newer car loses value fastest.

What the repair buysYour figures

The repair, per mile it buys

over miles

Buying the other car, per mile

of value gone, before any fuel

Start with three things: the quoted repair, how much longer you think the car will serve once it's done, and how far you drive in a year.

The same months, priced both ways
Over this windowRepair itReplace it
The bill nowrepair / purchase
Loan interest
Selling this car as-iscash in, only if you replace
Fuel or chargingper mile
Insuranceper year
Upkeepper year
Registration & feesper year
What you'd still own at the endvalue left, subtracted
Net cost
Per mile
Per month

If the repair buys less time, or more

months

months — yours

months

The bill does not change; the miles it is divided by do. This is the only figure on the page that rests on no forecast at all — a repair bill is a known number, and so are the miles you drive in a year.

The repair against what the car is worth
the rule of thumb — see below
The gap between the two paths
a mile

Anything left blank counts as zero. A blank insurance or upkeep field doesn't make those costs go away, and leaving one side's costs out while filling the other's in will decide the answer for you. The two end values are estimates, not arithmetic — the replacement's especially, since a newer car loses value fastest.

Informational only, not professional advice. This tool prices two paths using figures you supply, including how much longer you think a repaired car will last, which nobody can know. It is not a mechanical opinion on whether the repair is worth doing, and it does not judge whether a car is safe to keep driving — a mechanic does that. Nothing you type is sent anywhere; the arithmetic runs in your browser.

How this is calculated

Two ways of having a working car for the next stretch, priced over the same months and the same miles, then subtracted. Nothing more complicated than that — the difficulty in this decision has never been the arithmetic.

miles          = miles per year × (months ÷ 12)

repair per mile = repair bill ÷ miles          ← the headline

KEEP IT
  outlay       = repair + fuel + insurance + upkeep + registration
  net          = outlay − what it's still worth at the end

REPLACE IT
  outlay       = price + loan interest − this car sold as-is
                 + fuel + insurance + upkeep + registration
  net          = outlay − what the replacement is worth at the end

the answer     = the smaller net

The fuel line on each side is the division the rest of this site is built on: $ per gallon ÷ miles per gallon, or $ per kWh ÷ miles per kWh. Both cars get their own, because they are not the same car — an old one usually drinks more, and a replacement may not burn gasoline at all. The interest figure is the interest over this window only, not over the whole loan; the Auto Loan Payment Calculator prints both.

The one rule this page is built to contradict

“Never spend more on a repair than the car is worth.” It is the most repeated sentence in this decision, and it compares two figures that answer different questions. What a car is worth is the price of selling it. What a repair costs is the price of keeping it. A car's market value is set by what someone else would pay to take on its remaining life — which is exactly the thing you are about to buy, cheaper, by fixing it.

The result panel prints that ratio anyway, labelled, because you will be told it by someone before this is over and it is better seen than argued with. In the worked example this site uses, the repair costs 1.33× the car's value and keeping it is still the cheaper path by more than $6,000 over three years. The rule is groping toward a real question — how much longer will it last? — and never asks it. This page asks it, puts it in a field, and then shows you how much the answer moves if your guess is wrong by half.

The as-is value, and the dollar that gets counted twice

What the broken car would fetch today appears in exactly one place: as cash coming in on the replace side. That cash only arrives if you actually sell it. Counting it on the keep side as well — as a “loss” you suffer by not selling — is the commonest way this comparison is got wrong, and it always tilts the answer toward buying. The same discipline governs the two end values: each side is charged for what it spends and credited for what it still owns at the end, once.

Why the headline is per mile, not a verdict

The repair bill divided by the miles it buys is the only figure on this page that rests on no forecast at all. A written estimate is a known number and your annual mileage is a known number, so the division is exact. Everything else — both end values, next year's upkeep, how long the car has left — is an estimate, and the further down the page a figure sits, the more of them it carries. That is why the sensitivity strip moves the months rather than the money: if the repair still looks cheap at half your estimate of its remaining life, your guess is not carrying the decision and you can stop refining it.

The sources

What a replacement costs to own. AAA has published Your Driving Costs annually since 1950. Its 2025 edition put the average cost of a new vehicle at $11,577 a year, of which $4,334 was depreciation alone — the largest single category, and larger than fuel, insurance and maintenance put together for most vehicles. That figure is why a repair bill so often wins this comparison: an old car has already taken its depreciation, and the bill in your hand is visible in a way that $361 a month of quiet value loss never is. AAA is quoted here for scale and for its category list; none of its numbers enter your arithmetic, because a 45-model national average is a benchmark and not your car.

Why there is no threshold table on this page. The federal government answers “when should a car be replaced?” with exactly such a table. Under the Federal Management Regulation, 41 CFR 102-34.270 sets minimum retention standards for government vehicles — sedans and station wagons at “3 years or 60,000 miles”, four- and six-wheel-drive vehicles at 6 years or 40,000 miles, with the note that the standards are “stated in both years and miles; use whichever occurs first.” The companion section defines these as the “minimum number of years in use or miles traveled at which an executive agency may replace” a vehicle. It is a floor for a fleet, not advice for a driver, and it exists because nobody can inspect several hundred thousand vehicles one at a time. You can inspect yours. You have an estimate in your hand, you know what else the mechanic found, and you know your own mileage — which is strictly more information than any age-and- mileage cutoff can use. That is the whole reason this page is a calculator and not a rule.

How normal this decision is. S&P Global Mobility's annual vehicle-age analysis put the average age of the 289 million light vehicles on US roads at 12.8 years in 2025 — 14.5 years for passenger cars, an eighth consecutive annual rise, against a scrappage rate of 4.5%. The average car on the road is well past the age at which this question starts getting asked, and most of the people asking it are keeping the car.

What this does not cover

Risk, which is the real reason people replace cars. Every figure here is a middle expectation. It cannot price the spread around it, and an old car's spread is wider: usually it costs what you entered, occasionally it strands you. Read the gap between the two paths as what you are being paid to carry that risk, and decide whether it is enough — that judgment is yours, and it depends on things a calculator cannot ask, like whether there is a second car in the driveway.

Whether the repair is a good repair, or the car is safe. This is a cost comparison and nothing else. It does not know whether the estimate is fair, whether the part will hold, or whether the car should be on the road. A mechanic answers those, and a second written estimate is the cheapest thing you can buy in this whole process.

Also excluded: the time value of money — a dollar spent today and a dollar spent in year three are treated as the same dollar, which slightly favours the path that spends later. Inflation on any figure, likewise: fuel, premiums and repair labour all move, and everything here is held flat. And your time — days without a car, and hours at a shop, are real costs on the keep side that this page does not count.

Last reviewed: August 2026

Frequently asked questions

Is it true you should never spend more on a repair than the car is worth?

It is the most repeated rule in this decision and it does not survive contact with arithmetic. What a car is worth is the price of selling it. What a repair costs is the price of keeping it. Those are answers to different questions, and the ratio between them tells you nothing about which path is cheaper. A $3,200 repair on a car worth $2,400 is 133% of its value — and if it buys three more years at 12,000 miles a year, it costs about nine cents a mile, while replacing that car with a $28,000 one costs roughly thirty cents a mile in lost value and interest before a drop of fuel goes in. The rule is not useless, though: it is a rough proxy for a real question it never asks out loud, which is how much longer the car will last. On a car near the end, a big bill genuinely is throwing money away — not because it exceeds the car's value, but because it buys very few miles. Ask the question the rule is fumbling for, and you get an answer with a number attached.

How am I supposed to know how much longer the car will last?

You are not, and that is why this page asks rather than assuming — but you can bound it better than you think. Start with what actually failed and whether the rest of the car is in the same condition: a water pump on an otherwise sound car at 130,000 miles is a different proposition from a water pump on a car that also needs struts, tires and an exhaust. Ask the mechanic what else they saw while it was on the lift, which is the single most useful question in this whole decision and costs nothing. Then use the sensitivity strip in the result: it shows the same bill spread over half and double the months you entered. If the repair still looks cheap at half your estimate, the guess is not carrying the decision and you can stop worrying about it. If the answer flips between the two, the guess IS the decision — and the honest move is to get a second opinion on the car's condition rather than a more precise calculator.

Doesn't buying newer just move the money — no repair bills, and a warranty?

It moves the money into the one category that dwarfs repairs, and into the one that never sends an invoice. AAA's annual Your Driving Costs study finds depreciation is the single largest cost of owning a new vehicle, averaging $4,334 a year in its 2025 edition, against $11,577 a year all-in. A warranty covers repairs; nothing covers the value the car sheds in the driveway. That is why an old car with a repair bill so often wins this comparison: it has already taken its depreciation, and the bill you are agonising over is the whole cost, arriving all at once and visibly, rather than quietly at $360 a month. Put both in the table and see. Two things do genuinely favour replacing, and they are in the tool: a thirsty old car burns real money every mile, and further repairs on a car that has started needing them are rarely a one-off — that is what the 'other upkeep to come' field is for.

What if the car keeps breaking down — doesn't unreliability count for something?

It counts for a great deal, and this tool cannot price it. Every figure here is a middle-of-the-road expectation: it prices the repair you have been quoted and the upkeep you expect, and it says nothing about the spread around that. An older car has a wider spread — most of the time it costs what you entered, and occasionally it strands you two states from home on a holiday weekend. A newer one has a narrower spread at a higher average. What that tail is worth is a judgment about your own life: a spare vehicle in the driveway, a job you can miss a morning of, and a repair fund all make it cheap, and none of them are numbers this page can ask you for. The useful way to read the result is as the price of that insurance. If keeping the car is $6,000 cheaper over three years, that is what you are being paid to carry the risk, and you can decide whether it is enough. If the two paths land within a few hundred dollars, the arithmetic has told you it is not the deciding factor, and reliability is.

Related tools