Knowing when to sell
"Is this repair worth it?" is the wrong question, because it has no denominator. The question that resolves is: what does another year of this car cost per month, and what does the car that replaces it cost per month?
People try to answer the keep-or-sell question by comparing a repair estimate to what the car is worth. It is an appealing comparison and it is close to useless — a $1,800 repair on a $4,000 car sounds absurd and is often the cheapest move available, because the alternative is not zero dollars. The alternative is another car.
So convert both sides to the same unit: dollars per month of transportation. Everything below uses round illustrative numbers we made up to make the arithmetic legible — plug in your own.
The two numbers
Cost of keeping, per month. The repair in front of you, plus a realistic allowance for what else will come up, divided by the months you honestly expect to keep driving it. Add the difference in insurance and fuel if it is meaningful. Do not add what you already spent — that money is gone whichever way you choose, and it is the single most common reason people make this decision badly.
Cost of replacing, per month. Not the payment. The payment is only part of it. Take what the next car actually costs you each month: the value it will shed while you own it, plus any interest, plus the change in insurance and registration, plus its own maintenance. Then subtract what your current car would sell for today, spread over the same months — that is money you only get by selling.
Three comparisons
1. The big repair on a car that will keep going
Worth about $4,000. Needs $1,800 of work. Sound otherwise, records exist, and you would expect two more years out of it — 24 months, with maybe $1,200 of other things along the way.
Fix it. This is not close, and it is the case people get wrong most often because $1,800 against $4,000 feels wrong. A repair on a car you already own buys months at a price no replacement can match — provided the rest of the car really does have two years in it.
2. The big repair on a car that is nearly out of road
Worth about $3,000. Needs $3,500 of work. Rust at the rocker panels, the other axle is weeping, and realistically you are buying nine months.
Sell it, and sell it before the repair. The two numbers are nearly level, which people read as "it's a wash" — but a wash that leaves you owning a nine-month car is not a wash. When the arithmetic ties, the tie goes to the option that does not put you back here next spring.
3. Death by small repairs
Worth about $5,500. Nothing catastrophic, but it has been in the shop four times in a year — an alternator, a wheel bearing, a coil pack, an exhaust section — and something new is always starting.
The money says keep it. Decide whether the money is the point. Four shop visits also cost four days off work, two rides home and a standing background anxiety about whether the car will start. That is a real cost and it is legitimate to pay $220 a month to be rid of it — just name it as what it is rather than pretending the arithmetic said so.
Five signals that have nothing to do with a repair estimate
The arithmetic above assumes the decision is about money. Sometimes the timing is set by something else entirely, and these are worth watching for because they let you sell on your schedule rather than on the car's.
One expensive thing on an old car is an event. Two, in the same year, from unrelated systems, is a pattern — the car has reached the age where everything is at the end of its design life at once. The first repair is usually worth it. The second one is the signal.
If the car is going to need work to pass, you are about to spend money on a car you were already ambivalent about, and a buyer will price the same looming deadline into their offer. Selling a few weeks before is worth meaningfully more than selling a few weeks after a failure.
A commute that tripled, a car seat that does not fit, a job that needs a bed. No amount of favorable arithmetic fixes a car that is the wrong shape, and people tend to endure this far longer than they endure a mechanical problem.
If a factory or purchased warranty is close to expiring, the car's risk profile changes on that date — for you and for a buyer. Whether that argues for selling or for using the remaining coverage on the things you have been putting off depends on the car, but it is a date worth knowing.
Demand for particular body styles moves with the season and with fuel prices. You cannot time this precisely and you should not try, but if you are already within a year of selling and your car is having a good month, that is not nothing.
Sell before the failure, not after
The most expensive version of this decision is the one made in a service waiting room. A car sold while it runs well, with records and an honest listing, is sold at retail to a private buyer. The same car sold three weeks later with a failed transmission is sold at scrap-adjacent money to whoever buys projects, and the gap between those two outcomes is far larger than the repair would have been.
Which is the real argument for doing the arithmetic before anything breaks. If the numbers say a car has a year left in it, you have a year to sell it properly — clean it, gather the records, list it carefully — instead of thirty minutes to make a decision you will explain to yourself later.
The sentence that costs people the most money: "I've already put so much into this car." Money spent is spent. It is identical under both options, so it belongs in neither column. The only figures that matter are the ones ahead of you — and the only test is which choice buys the next twelve months more cheaply, counting your time as worth something.
Every figure on this page is a round illustrative number we computed for the example, not an average, a survey result or a measured repair cost. Depreciation, insurance, interest and repair pricing vary enormously by vehicle, condition and circumstance; use the structure, not the numbers.