Signs a Car Has Been Repainted: Paint Depth and Panel Gaps

A paint depth gauge does not measure paint. Read the standard that governs it and the claim gets narrower than the box it came in: ASTM D7091-22 says these instruments "use a sensing probe or magnet to measure the gap (distance) between the base metal and the probe," and that distance is what the display reports as coating thickness. Body filler sits in that gap. So does primer, so does a vinyl wrap, so does the clearcoat an assembly plant sprayed in 2016. The number is real. It is just not a number about paint.

That distinction sets the value of a five-minute walk around a car: worth something, not worth a verdict. I check documents rather than cars and I am not a technician. Nothing below concludes that a car was wrecked. It produces questions with paperwork behind them.

The gauge is measuring distance to steel

ASTM D7091-22, the current designation, is titled "Standard Practice for Nondestructive Measurement of Dry Film Thickness of Nonmagnetic Coatings Applied to Ferrous Metals and Nonmagnetic, Nonconductive Coatings Applied to Non-Ferrous Metals." The title is the whole limitation. Magnetic gauges work over steel. Eddy current gauges work over non-ferrous metal, which on a car means aluminium. Section 1.2 adds that the procedures do not apply to coatings that deform under the probe, because the probe has to sit directly on the surface.

None of this is in the box copy:

• An aluminium hood on a steel car needs the other mode. A gauge that only does ferrous will report nonsense or nothing on an aluminium bonnet, boot lid or fender, and plenty of cars from the last fifteen years mix the two. A dual Fe/NFe gauge switches automatically and usually tells you which substrate it found, which is itself information.
• Plastic is invisible to both. I-CAR's article of 26 April 2024 on plastic film gauges says the only way to confirm a maximum film thickness on a bumper cover is "a film gauge capable of accurately measuring paint film on a plastic substrate," and that some manufacturers now require or recommend one in their certified collision programs. A common Fe/NFe gauge is blind on every bumper cover, mirror cap and spoiler on the car.
• A high reading means non-metallic buildup, not paint. Filler under a repaint reads as coating. So does a thick ceramic layer, and so does a paint protection film. A big number tells you there is more material between the probe and the metal than the plant put there, and stops.

That leaves the question of what "normal" is. The published sources disagree, which is the most useful fact on this page.

All three read 27 September 2026. Two of them converge on a strip-it threshold that is the same number in different units: Axalta says film builds of 12 mils or more, not counting polyester fillers, should be stripped before any repair, and PPG puts the sand-back-or-remove line at 300 µm or greater. Twelve mils is 305 µm. That agreement is worth more than the disagreement above it, because it tells you where readings stop being ambiguous.

Before citing any of them at a seller, read the fine print. The Axalta sheet carries its own disclaimer, in capitals, that the information is "ABSOLUTELY NOT BINDING" and is provided as-is. And all three are paint and training organisations describing industry averages, not the manufacturer of the car in front of you publishing a specification for that panel. No such per-panel specification surfaced anywhere public in this search, which is why the next section matters more than this one.

Compare a panel to its twin, not to a chart

The strong signal is spread, and you already have a reference sample: the same car.

Take four to six readings per panel rather than one, spaced out — high near the beltline, middle, low near the rocker, and one a hand's width in from an edge. Write them down. The pairs that matter are the ones where the factory did the same thing twice: left front fender against right front fender, driver's door against passenger's door, rear door against rear door, quarter panel against quarter panel. Then compare each door to the door behind it on the same side.

The roof and the rear quarter panels are the closest thing to a factory baseline on a specific car, because they are welded rather than bolted, so a repair done to a price has little reason to reach them. If the roof and both quarters cluster tightly and one front fender sits well above them, you have found the interesting panel. If everything on the car sits high and tight together, you are probably looking at a whole-car respray, which is a different conversation and not automatically a bad one.

What the patterns do and do not support:

A single panel well above its mirror twin. Supports: that panel has more non-metallic material on it than the factory put there. Does not support: a collision, a severity, or a date. Ask what the panel is, and whether it is a panel people hit.

Readings that climb from a few hundred microns past a thousand across a few inches of one spot. Non-metallic buildup with a shape and a centre, which you can map with the probe. Filler behaves like that, and so does primer piled over a repair. What put the dent there is not in the reading.

Everything uniform and inside every published range. This is the pattern that fools people. A panel replacement using a pre-primed OEM panel, refinished by a shop that measures its own film build, can land inside those ranges on purpose. A normal gauge reading is not an all-clear, and treating it as one is the same mistake as reading a clean title as "nothing happened."

Sight down the side, then feel the edges

The most quotable public source on the visual half is a field guide to detecting collision repair hosted by the Virginia Motor Vehicle Dealer Board at mvdb.virginia.gov, read 27 September 2026. Read it knowing what it is: produced by Stop Curbstoning, founded by AutoTec, LLC of Birmingham, Alabama, it recommends a commercial history report by name and carries no date anywhere in its two pages. Industry material on a state regulator's server, not agency guidance — and still more specific than most of what a search returns. Its own premise is the honest one: now that computers assist "with everything from frame straightening to color matching, it's a lot harder to tell if a vehicle has been in a collision."

On reflections, which cost nothing: "factory paintwork is mirror-smooth, even when dirty or dinged up. Signs of repair often reveal themselves in a slightly rippled reflection, or a reflection that doesn't exactly line up across two body panels." If the car is too dirty to throw a reflection, ask for it washed, and notice whether that request is a problem.

On gaps, the guide gives a method rather than a number: while sighting down the sides, "check to make sure the body panels all fit flush, and that the gaps between the panels are even from top-to-bottom and side-to-side."

I went looking for a published tolerance to put against that and came back empty. Trade press and machine-vision vendors discuss production gap and flush control in the region of a millimetre, but no manufacturer document I could reach states an allowable gap for a specific model, and the one article that looked like it carried figures answered a direct request with a 403 on 27 September 2026. So there is no number here, and an article that hands you one without naming the model and the source is guessing. The test is symmetry: photograph the bonnet-to-fender gap on both sides from the same distance and angle, and look for a gap that tapers on one side and stays parallel on the other. Doors that need a shove to latch on one side and shut with a touch on the other belong in the same list.

Then the part nobody does, from the same guide's hands-on section: "Feel around the painted edges of the hood, trunk, and doors. It is very difficult — almost impossible, in fact — to repaint a body panel and get the edge as smooth as a factory-finish." Their instruction is to calibrate yourself first: "try this on a car you know to never have been repainted so you know what a factory finish feels like." Ninety seconds on a low-mileage car in somebody's driveway does that job.

Overspray hides in five places and one of them is a wheel well

The same guide lists where excess paint collects when a panel was sprayed in place rather than off the car. Quoted as written: "Around electrical conduits in door jambs. Near door, hood, and trunk hinges. On window and door seals. Around seals and fittings in the engine compartment firewall. Along the trim pieces inside wheel wells."

Wheel wells are worth the crouch. They are awkward, they are dirty, and a shop working to a price does not mask them well.

Two more from the same document that cut against intuition. Peeling clearcoat in small patches, especially in corners and hard-to-reach areas, "may indicate a poorly prepped paint job" — different from the broad sun damage that shows up on bonnets and roofs. And: "beware of too-perfect paint. If the chin fascia beneath the front bumper is pristine, it may have been recently replaced or repainted."

Under the car, without going far under, the guide names weld seams, overspray, excess undercoating and "holes indicating missing bolts or parts," then adds the comparison worth copying: "Look at the bumper mounts — if they look newer than the rest of the underbody components, that's a red flag." A torch and a hand mirror are the whole tool list.

A word on fasteners, because the internet is confident about them and the confidence outruns the evidence. A bolt that has been turned shows tool marks on its flats, and that much is observable. Whether it means anything depends on whether that fastener is meant to be adjusted, and plenty are: hinges get adjusted, bumper bars come off for towing eyes, fenders come off to fix a wing mirror. Tool marks are a question about that fastener, not a finding about the car.

The labels the rules force to carry a date

A gauge cannot do this part. Several federal rules require markings that carry a date or an identity, and all of them are readable in a parking lot. All CFR text below is the text in force on 1 September 2026, pulled from the eCFR versioner API on 27 September 2026.

The certification label, on the driver's door area. 49 CFR 567.4(b) says the label "shall be riveted or permanently affixed in such a manner that it cannot be removed without destroying or defacing it," and 567.4(c) puts it on the hinge pillar, the door-latch post or the door edge that meets it, next to the driver's seating position. Paragraph (g)(2) requires the "Month and year of manufacture," which it allows as "June 2000" or "6/00." Paragraph (g)(3) requires the GVWR in pounds. That label is your reference date for everything that follows, and a label that looks newer than the jamb around it is its own question.

The tire and loading placard, a separate label. FMVSS 110, S4.3 requires a placard "permanently affixed to the driver's side B-pillar," with fallback positions if that is not legible and prominent. Item (d) is the original tire size "for the tires installed at the time of the first purchase for purposes other than resale." Two labels in the same doorway in visibly different condition is a cheap observation.

Tires. 49 CFR 574.5(b)(3) requires the last four digits of the tire identification number to be the week and year of manufacture, and gives its own worked example: "0109 means the tire was manufactured in the first full calendar week of 2009." Four tires from the same week were fitted together. One tire from a different year was fitted alone.

Seat belts. FMVSS 209, S4.1(j): "Each seat belt assembly shall be permanently and legibly marked or labeled with year of manufacture, model, and name or trademark of manufacturer or distributor." A belt whose year is later than the car's build year was replaced, and belts are replaced after they lock up in a collision.

Glass. FMVSS 205, S6.2 requires a prime glazing manufacturer to add, to the marks already required by ANSI/SAE Z26.1-1996, "the symbol 'DOT' and a manufacturer's code mark that NHTSA assigns to the manufacturer." Note what is not in there: a date code. The month-dot convention people photograph is manufacturer practice, not a federal requirement, so do not build an argument on it. What the rule does give you is a maker's code on every pane, and original glass usually carries one supplier's code throughout — so a single pane coded differently from its neighbours is the pane to ask about. S5.1.2 puts aftermarket replacement glazing under the same standard, so a new pane is marked too — just by somebody else.

The dash VIN. 49 CFR 565.13(e) requires the VIN to appear "clearly and indelibly" on a part "that is not designed to be removed except for repair," or on a plate permanently affixed to such a part, and 565.13(f) requires it to be readable through the glazing from outside next to the left windshield pillar with characters at least 4 mm high. Read it, photograph it, and match it to the title and the jamb label before anything else on this page.

None of these is decisive alone. Tires wear out, windscreens meet stones, a belt gets swapped under a recall. What a buyer is looking for is a cluster: three or four dates that group in one month two years after the build date, on one corner of the car.

Whether the panels are marked at all depends on a weight and a list

One more marking is the only one that separates a replaced panel from a repainted one, and it reaches fewer cars than people assume. Finding out starts with the label you already read.

49 CFR 541.3(a) applies the Federal Motor Vehicle Theft Prevention Standard to passenger cars, to multipurpose passenger vehicles "with a gross vehicle weight rating of 6,000 pounds or less," and to light-duty trucks of 6,000 lb GVWR or less that NHTSA determined to be high theft. So the GVWR on the certification label is the first gate. A full-size pickup or a large SUV rated above 6,000 lb is outside the standard entirely, and there is nothing to look for. Paragraph (b) removes more: manufacturers who sell fewer than 5,000 vehicles a year in the US, and any line with annual production of not more than 3,500.

If the car clears those, 541.5(a) lists eighteen parts that carry a number when present — engine and transmission, both front fenders, the hood, all four doors, both bumpers, both rear quarter panels on a car, the side assemblies on an MPV, the cargo box on a light truck, and the decklid, tailgate or hatchback. Paragraph (b)(1) makes that number the VIN.

What turns that into a usable check rather than a trivia question:

• A replacement panel is marked differently on purpose. 541.6 requires the replacement part maker's registered trademark or other unique identifier plus the letter "R," at least one centimetre high, and paragraph (f) adds the symbol "DOT" at the same height within five centimetres of it. Paragraph (b) is the clincher: "A replacement part subject to paragraph (a) of this section shall not be marked pursuant to § 541.5." VIN means original. Trademark plus R plus DOT means replacement.
• Peeling the label off leaves evidence, by design. 541.5(d)(1)(v) requires removal to destroy the label and to "discernibly alter the appearance of that area of the part where the label was affixed by leaving residual parts of the label or adhesive in that area, so that investigators will have evidence that a label was originally present." A clean rectangle of adhesive shadow on a door shut face is the finding.
• You may not find it even when it is there. The label goes on an interior surface where practicable, inside a target area the manufacturer designates, and 541.5(e)(2) lets that target area information be filed with NHTSA under a request for confidential treatment. Absence of a label you looked for is not absence of a label.

And then the part that deflates the whole check for most used cars. Appendix A-I to Part 541, last amended at 88 FR 77227 on 9 November 2023, lists the lines exempted from parts marking under 49 CFR part 543 because they carry antitheft devices. Nineteen manufacturers appear, and the lines read like the American used car market: Camry, Corolla, RAV4, Civic, Accord, CR-V, Escape, Explorer, Equinox, Cruze, Altima, Rogue, Forester, Outback, Jetta, Golf, Jeep Grand Cherokee, Jeep Wrangler JL, 3 Series. The most common cars you will look at are the ones with nothing to check.

Do not read the gaps in that list as a pattern. The Malibu and the Traverse are absent while the Impala and the Cruze are on it, and 543.5(a) is why the list looks patchy: a manufacturer may petition for an exemption of one car line per model year, so a line is missing because nobody obtained an exemption for it. Absence is not a statement about the car — and it does not mean the panels are marked either, because the line still has to clear the weight and volume gates in 541.3. The appendix also dates itself badly. Footnote 1 marks the lines exempted beginning with MY 2021, but most entries carry no start year at all, so an older example of a listed line may have been marked before the exemption was granted and the appendix will not tell you when.

One worked example, because the two documents do connect. NHTSA's own sample partial VIN in the vPIC documentation, 5UXWX7C5*BA, decodes through the vPIC API to a 2011 BMW X3, body class "Sport Utility Vehicle [SUV]/Multipurpose Vehicle [MPV]," GVWR returned as "Class 1D: 5,001 - 6,000 lb," queried 27 September 2026. Under 6,000 lb, so Part 541 could reach it — except "X3 (MPV)" sits in Appendix A-I under BMW. No marks to hunt for. Note that vPIC returns GVWR as a class band rather than a figure, so checking it against the pounds printed on the jamb label catches a gross mismatch and nothing finer. What vPIC and the recall lookup each tell you is the prerequisite to all of this.

What the five minutes is actually for

It buys a list, and the list has two destinations.

The seller gets the questions, each with a document behind it: why is this fender 240 µm when the other one is 140, whose repair order covers the replaced offside door glass, what happened in the month those three dates cluster in. The Virginia-hosted guide gets the framing right: "the biggest red flag is a disparity between what the seller says and what the car shows. An owner knows his or her car's history. A curbstoner doesn't."

The person you pay gets the same list in writing, before anyone lifts the car: name the panels whose readings were out of family, name the corner where the dates cluster, ask for structural inspection of that area specifically. Scope is set when you book, which is the whole argument in how to hire an independent pre-purchase inspection. Sorting what comes back into numbers, questions and exits is its own job — negotiating points versus walk-aways. And if the readings and the smell both point at the floor rather than the fenders, water leaves evidence in a different set of places entirely.

The FTC's own page on buying a used car from a dealer, read 27 September 2026, puts the boundary in one line: "A vehicle history report is not a substitute for an independent vehicle inspection." A paint gauge is not a substitute either. The same page answers the objection you will hear — if the dealer will not let the car leave the lot, "you may be able to find a mobile inspection service," which brings the lift's worth of scrutiny to the lot instead. A gauge and a torch get you as far as knowing which panel to name when you book it.