Tesla spent two years and change arguing that its headlights weren’t too bright. It lost, and now roughly 20,000 owners are waiting to find out what the fix even is.
The 20,349-vehicle recall that showed up in NHTSA’s database this week isn’t a surprise development. It’s the endgame of a regulatory fight that started in March 2024, and the paperwork behind it is far more interesting than the headline number.
What’s actually wrong
Here’s the part most coverage skips: this isn’t about the headlights being blindingly bright where you look. It’s about a very specific slice of sky.
Federal Motor Vehicle Safety Standard No. 108 sets photometric limits at dozens of test points around a headlamp’s beam pattern. The one that matters here covers the zone from 10 degrees above horizontal all the way to 90 degrees up, spanning 90 degrees left to 90 degrees right. Per the Federal Register notice documenting the case, the maximum allowed intensity in that zone is 125 candela.
Tesla’s affected lamps measured as high as 230.1 cd. That’s 105.1 cd over the limit — roughly 84% too much light thrown into the region above the beam cutoff.
For scale, one candela is approximately the output of a single wax candle. We’re talking about the difference between two candles’ worth of stray light and roughly four. Which sounds trivial, and that was precisely Tesla’s argument.
The testing data is worth knowing too. Tesla’s supplier, Marelli Automotive Lighting, sampled 25 right-hand and 25 left-hand lamps. Right-hand units came in between 136.2 and 230.1 cd. Left-hand units ran 117.5 to 160.3 cd. A Transport Canada-recognized lab independently measured a left-hand lamp at 171.27 cd. Note what that spread implies: essentially every right-hand lamp in the sample was over the 125 cd ceiling.
Why “off the road” wasn’t good enough
Tesla’s position was that the noncompliant light sits high and outboard — 30 degrees inboard and 20 degrees up within the zone — putting it outside anyone’s natural line of sight. Light aimed at the treetops can’t blind an oncoming driver, so where’s the harm?
The company backed this with track testing, running one Model 3 and one Model Y through the Adaptive Driving Beam protocol from FMVSS 108, Scenario #1, at 60 mph in an opposite-direction pass. The vehicles met the ADB illuminance limits despite the noncompliant lamps.
NHTSA’s answer, and this is the technically important bit, is that Tesla misunderstood what those upper test points are for. They don’t exist to prevent direct glare. They exist to limit veiling glare — light that scatters off water droplets, ice crystals, and particulates and bounces back as a luminous haze that washes out contrast. The agency’s notice cites the Illuminating Engineering Society definition: a reflection superimposed on what you’re trying to see, obscuring detail by killing contrast.
If you’ve driven a car with poorly controlled optics through heavy fog, you know exactly what this is. You turn the high beams off because they make things worse, not better. That’s stray light above the cutoff being reflected straight back at your retinas.
Tesla’s ADB testing was conducted on a straight, dry road under a clear sky — one of eight scenarios in the standard. A commenter on the docket called Tesla’s arguments “not relevant, incomplete and unconvincing,” pointing out that curve scenarios could be more severe and that ADB detector positions were never designed to catch veiling glare. NHTSA agreed with essentially all of it.
Then there’s the footnote that quietly guts the whole ADB argument: the standard also requires ADB lamps to pass laboratory testing under S9.4.1.6.4.3, and the subject lamp fails that too.
The precedent Tesla tried to dodge
Tesla’s petition leaned hard on distinguishing itself from a 2022 GM denial involving replaceable-bulb low beams that measured 450 to 470 cd — more than triple the limit. Tesla’s argument was essentially one of degree: GM was off by 3.6x, we’re off by 1.8x, treat us differently.
NHTSA didn’t bite. The agency found the two cases structurally identical — same zone, same failure mode — and said the magnitude and location of the exceedance didn’t change the fact that Tesla never produced evidence addressing veiling glare in adverse weather. Degree of violation doesn’t create a safe harbor when nobody tested the actual risk pathway.
There’s history here beyond GM, too. In 2022 the agency denied a Soft Lights Foundation petition that argued LED headlamps were inherently noncompliant, ruling that LED technology isn’t automatically a problem — the lamp has to meet photometry, not the light source. NHTSA cited that decision here, noting the same petition had specifically complained about glare from Model 3s.
Why there’s no fix yet
Tesla hasn’t finalized a remedy, and if you understand the failure mode you can see why that’s a genuinely hard problem rather than corporate foot-dragging.
This is optics. The stray light is a function of lens geometry, reflector surface, and how the LED source couples into the projector — physical hardware that was built wrong at the supplier. It is not a calibration value sitting in a config file. Tesla’s usual party trick, the over-the-air update, has nothing to grab onto here.
Which leaves a short list of unappealing options: replace the headlamp assemblies outright on 20,000-odd cars, or engineer some form of internal masking or shield retrofit that kills output in that upper zone without wrecking the compliant portion of the beam pattern. Either way, somebody is physically touching every one of these vehicles.
The good news for owners is that 49 U.S.C. 30120 obligates Tesla to provide that remedy at no cost. If you’ve already paid out of pocket for a headlamp replacement on an affected car, the same statute provides for reimbursement — keep the invoice.
What owners and buyers should actually do
Model 3s from 2017 through 2023 and Model Ys from 2020 through 2023 are in scope. Run your VIN through NHTSA’s lookup rather than assuming, because the affected population was identified through Tesla’s build and service records, not a clean production-date cutoff.
Two practical notes that go beyond the recall itself:
Get your headlamp aim checked while you’re at it. Vertical misaim stacks directly on top of a photometry exceedance — a lamp that’s 84% over in the upper zone and aimed a degree high is a materially worse offender than either problem alone. Aim drifts after front-end collision repair, suspension work, or sustained heavy rear loading, and almost nobody checks it.
And if you’re shopping used: an unremedied recall follows the VIN, not the owner. Federal law bars dealers from delivering new vehicles with open recalls, but there’s no equivalent federal prohibition on used sales. A private-party 2021 Model Y with this campaign outstanding is entirely legal to buy — you just inherit the wait, and the paperwork.
One last thought worth sitting with. Tesla told regulators it had received no complaints, no accidents, no injuries tied to these lamps. NHTSA’s response was that the absence of complaints proves nothing, because nobody dazzled on a rainy night at 60 mph pulls over to identify the make and model of the car that did it. That reasoning now sits in the Federal Register, and it applies to considerably more than 20,000 Teslas.