Repair or Replace: The Economics, Worked Out
The short answer
- When damage qualifies for repair, repair is almost always the correct first move, because the downside is capped at the repair fee and the upside avoids a much larger bill.
- The break-even probability is low. On a camera equipped car, a $100 repair only has to work about 8 percent of the time to pay for itself.
- Replacement later usually costs the same as replacement now, so a failed repair costs you the repair fee, not the whole decision.
- Go straight to replacement when the damage fails the repairability rules: it reaches the edge, the inner ply is broken, or the break is too large. Damage in the driver's primary viewing area is a trade off, not an automatic replacement.
- If your deductible is $500 or $1,000 and your quote is near it, you are effectively paying cash either way, so treat the claim as irrelevant to this decision.
The repair versus replace question is not a taste question, it is an expected value calculation with four inputs: what repair costs now, how likely it is to hold, what replacement costs if you end up there anyway, and how long you intend to keep the car. Run those numbers on almost any repairable break and repair wins by a wide margin, because the loss if it fails is one repair fee while the saving if it holds is several hundred dollars or more. This page shows the arithmetic so you can run it on your own damage.
The model in one line
Compare two paths. Path A is replace now. Path B is repair now, and replace later only if the repair does not hold.
- Cost of Path A = R, your out of pocket replacement cost today.
- Cost of Path B = C + (1 minus p) times R', where C is the repair fee, p is the probability the repair holds for the rest of your ownership, and R' is the replacement cost if you need it later.
Repair is the better choice whenever C + (1 minus p) times R' is less than R. Because R' is essentially the same as R (windshield prices do not fall as your car ages, and calibration requirements do not go away), that simplifies to a clean rule: repair wins whenever p is greater than C divided by R.
The break-even probability. C divided by R is the only threshold you need. A $100 repair against a $600 replacement breaks even at p = 0.167. Against a $1,200 replacement it breaks even at p = 0.083. In other words, on a camera equipped vehicle a repair only needs to work about one time in twelve to be worth attempting. Realistic hold rates for correctly performed repairs on qualifying damage are far above that.
Two refinements make it honest. First, a failed repair is not usually a total loss, because many shops credit the repair fee toward the replacement, which lowers the effective C. Second, a repair on a car you will only own another year does not need to hold forever, only until you sell, which raises p. Both refinements push in the same direction: toward repairing.
Getting your four inputs right
| Input | How to get it | Typical 2026 range |
|---|---|---|
| C, repair cost | Single chip, professional, cash | $60 to $150, plus $10 to $40 per extra chip |
| C, long crack repair | Where a shop will attempt it | $80 to $200 |
| C, DIY kit | Retail kit, one break | $10 to $30 |
| R, replacement, older car, no sensors | Aftermarket glass, no calibration | $250 to $450 |
| R, mainstream car with rain sensor and camera | Aftermarket glass, before calibration | $450 to $900 |
| R, luxury with a head up display | Aftermarket or dealer equivalent glass, before any OEM uplift | $900 to $1,800 |
| R, luxury, HUD and an OEM sourced part | Often dealer supplied. The OEM uplift is 30 to 70 percent on glass and labor | $900 to $2,000 or more |
| R, EV or low volume model | Large raked glass, sensor cluster, thin aftermarket supply | $1,000 to $2,200 |
| Calibration to add to R | Only if camera equipped | Dynamic $150 to $350. Static $250 to $600. Both procedures together, $400 to $800. |
| p, probability repair holds | Judge from the damage, see below | Use the bands in the next table |
| Deductible | Your comprehensive policy | Commonly $250, $500 or $1,000 |
Get R from the cost estimator and check whether your damage even qualifies with the repair or replace tool. The variables that set R are broken down in what drives the price.
Estimating p without pretending to be precise
Nobody can hand you a number for p on your specific chip. What you can do is place your damage into a band, then check whether the band clears your break-even threshold, which is usually only 8 to 20 percent. Almost every band clears it.
| Damage | Working assumption for p | Why |
|---|---|---|
| Fresh bullseye or half moon under a quarter, away from the edge | Very high | Single clean cavity, easy to evacuate and fill |
| Fresh star break with short legs, away from the edge | High | Legs fill well when the break is clean and the tech runs proper vacuum cycles |
| Combination break, weeks old, some dirt in the pit | Moderate | Contamination limits both fill and clarity |
| Long crack, 6 to 12 inches, not at the edge | Moderate | Repairable by some shops, more likely to run later |
| Any damage within about two inches of the edge | Low | Edge stress concentrates there, see edge cracks |
| Damage over about 1.5 inches across, or through the inner ply | Not repairable | Outside the repair standard entirely |
The technical rules behind those bands are in can my windshield be repaired. Note that p falls with time, not with mileage. Dirt and water enter a break within days, which is why repairing quickly is worth more than shopping around for a $20 discount.
Scenario 1: 2019 crossover, camera, $500 deductible
A fresh star break the size of a nickel, in the passenger half, six inches from any edge. Repair quoted at $110. Replacement quoted at $720 for the glass and install, plus a $300 dynamic calibration, so R = $1,020. The comprehensive deductible is $500, and this state does not mandate a waived deductible for glass.
- Path A, replace now: cash $1,020, or file a claim and pay the $500 deductible.
- Break-even p: C divided by R = 110 / 1020 = 0.108. The repair needs to hold about 11 percent of the time to be worth doing against the cash price.
- Against the deductible: if you would claim instead, R becomes $500 and break-even p = 110 / 500 = 0.22. Still low.
- Expected cost of Path B at a conservative p = 0.85: 110 + 0.15 times 1020 = $263. Against $1,020 to replace now.
Decision: repair. It is not close. And in many states, and under many policies, chip repair is covered with the deductible waived entirely, which drops C to zero and makes the decision automatic. Check whether your insurance covers repair before you pay cash.
Scenario 2: 2007 sedan, no sensors, $1,000 deductible
A 10 inch crack running in from the lower passenger corner, three weeks old. Crack repair quoted at $150. Replacement quoted at $340, aftermarket glass, no calibration. The $1,000 deductible is above the replacement cost, so insurance is irrelevant: this is a cash decision either way, which is exactly the situation described in cash or claim.
- Break-even p: 150 / 340 = 0.44. The crack repair now has to hold nearly half the time just to break even.
- Actual p is poor here: the crack starts at the edge, which is the highest stress region of the glass, and it is three weeks old and contaminated.
- Expected cost of Path B at an optimistic p = 0.5: 150 + 0.5 times 340 = $320, barely under $340, and you still have a visible crack line and a second appointment ahead of you.
Decision: replace. This is the classic case where the model flips. The repair is expensive relative to replacement, the replacement is cheap because there is no sensor stack and no calibration, and p is genuinely low because of the edge origin. Cheap replacement is what makes cheap repair pointless.
Scenario 3: 2023 EV with HUD, two chips, keeping the car five years
Two small bullseyes low on the passenger side, both fresh. Repair quoted at $130 for the first and $30 for the second, so C = $160. Replacement is $1,850 for the glass and install (HUD wedge, acoustic, IR coated, dealer only part) plus $550 static calibration, so R = $2,400.
- Break-even p: 160 / 2400 = 0.067. Under 7 percent.
- Expected cost of Path B at p = 0.9: 160 + 0.1 times 2400 = $400.
- Long horizon effect: keeping the car five years slightly lowers p, because there is more time for a marginal repair to move. It does not come close to changing the answer at a 6.7 percent threshold.
- Second order benefit: repairing both chips removes two future crack origins on a windshield that costs $2,400 to replace. That is the cheapest risk reduction available on this car.
Decision: repair, immediately, and do not wait for the weekend. The more expensive your glass, the more aggressively you should repair, because R is the number the threshold is divided by. This is the counterintuitive result people get wrong: owners of expensive cars often assume repair is beneath their vehicle, when in fact they have the strongest financial case for it of anyone.
Why the asymmetry runs one way
Three structural facts make repair the default when damage qualifies.
- The downside is capped and small. The worst realistic outcome of an attempted repair is that it does not stop the break and you pay for the replacement you were going to pay for anyway, minus any credit. A repair does not make the windshield harder or more expensive to replace, and it does not damage anything else on the car.
- The upside is the entire replacement cost. On a sensor equipped vehicle that is $600 to $3,000 delivered, calibration included.
- The replacement price does not decay. Waiting does not make R cheaper, so deferring the replacement costs you nothing in the case where you eventually need it. There is no time value penalty pushing you toward replacing now.
The asymmetry only reverses when C rises toward R, which happens on cheap glass, or when p collapses, which happens with edge damage, large damage, contaminated old damage or damage in the driver's primary sight line where a cosmetically imperfect repair is itself a problem. Those two conditions are the whole of the replace-first case.
When to skip repair entirely
| Condition | Why repair is off the table |
|---|---|
| Damage penetrates the inner glass ply | Repair only addresses the outer ply. Nothing can be done from the pit side. |
| Crack reaches or starts at the edge | Edge stress and the bonded perimeter make continued spreading likely, see will a crack spread. |
| Damage larger than about 1.5 inches across, or four or more separate breaks | Outside the accepted repair scope. Between 1 and 1.5 inches is borderline and technician dependent. |
| Contamination or long standing moisture, plus a low R | Low p and cheap replacement together flip the arithmetic. |
| Existing pitting and glare across the glass | You are replacing for optics anyway, see pitting and glare. |
| Damage over a camera or sensor viewing zone | Resin in the optical path can interfere with the camera, and calibration may not pass. |
Damage in the driver's primary viewing area is deliberately not on that list. It is the wiper swept band in front of you, below the AS-1 line, and it is a trade off rather than a hard stop. Repair standards discourage it and plenty of shops decline it, because even a good repair leaves a small blemish and slight distortion exactly where you look. But the arithmetic does not disappear: repairing stops the break spreading now, costs $60 to $150, and leaves you free to replace the glass on your own schedule instead of on the break's. A driver who would rather live with a faint mark than pay for glass today is making a defensible choice, and a driver who cannot tolerate a mark in the sight line is making an equally defensible one. What settles it is how much the blemish will bother you and whether your state inspects glazing, not a rule. See the AS-1 line and driver view.
Where the deductible fits
Insurance changes R, not p, so it only ever changes the threshold, never the shape of the decision. Three positions matter.
- Deductible above the replacement quote (common with a $1,000 deductible on an older car): the claim is worthless, you are a cash buyer, use the raw quote as R.
- Deductible below the replacement quote ($250 or $500 on a modern car): your effective R is the deductible, which raises the break-even p, but it is still typically only 20 to 40 percent. Repair still wins on qualifying damage.
- Repair covered with the deductible waived: C is effectively zero and there is no calculation left to do. Many carriers do this because a $100 repair is cheaper for them than a $1,000 replacement, and some states require it for glass. See zero deductible glass states.
What insurance does not change is the physics. A claim will not make an edge crack repairable, and a waived deductible is not a reason to attempt a repair that is going to fail. If a repair does fail, read failed repair and redo before you accept a second attempt on the same break.
Frequently asked questions
Does a repair reduce what my windshield is worth later?
No, because a windshield has no resale value of its own. A repaired chip is not a defect that lowers a car's trade-in value in any measurable way. The only asset at risk is the glass you already own, and repair extends its life rather than shortening it.
If the repair fails, do I lose the money I spent on it?
Usually you lose part of it. Many shops credit the repair fee toward a replacement they perform, and most insurers that waived a deductible for repair will still handle the replacement normally. Ask about the credit before the repair, not after.
I plan to sell the car in six months. Repair or replace?
Repair, if it qualifies. A short ownership horizon makes the cheap, fast option better, and a properly repaired chip does not fail a sale. Replacement only becomes right if the damage blocks the driver's view or would fail a state inspection at transfer.
Is it worth repairing two or three chips at once?
Yes. Additional chips are typically $10 to $40 each on the same visit against $60 to $150 for the first, so the marginal cost is small. Multiple repairs also remove multiple future crack starting points at one sitting.