How Do You Find Where a Windshield Is Leaking?
The short answer
- Use a garden hose with no nozzle: GM floods the suspect area for up to 5 minutes, Nissan for at least 2, and Daimler holds each spot at least 1 minute so hidden cavities can fill.
- Prove the leak with water falling from the top down, as Volkswagen does, then locate it from the bottom up, one zone at a time, so each new drip belongs to the zone you just wetted.
- Never use a pressure washer. GM says its seals are designed for natural elements and "cannot take into consideration" high pressure car washes.
- To pin the exact gap, run the blower on high with recirculation off and doors shut, then spray the outside seam with at least a quarter cup of soap per gallon of water and look for bubbles.
- Check the body before the glass: GM, Ford, Nissan and Daimler bulletins traced "windshield" leaks to roof seams, body joints and the cowl.
- A gap in the bond itself means the glass comes out; a body seam or moulding gap is sealed in place, and every fix is re-tested after the sealer cures (at least 1 hour for Ford's urethane reseal).
Flood one small zone at a time with a garden hose that has no nozzle on it, for 1 to 5 minutes per spot, with a helper inside watching, working from the bottom of the glass upward. When water shows up, push air out from inside the car and brush soapy water over the outside seam: the bubbles mark the gap. That is the method in the carmaker service documents gathered below, and the same documents show the other half of the answer: a good share of leaks that look like the windshield turn out to be a roof seam, a body joint or the cowl.
How the carmakers test for a windshield leak
Every procedure below uses low pressure water, a person inside, and a set time per spot. They differ on the time and on where the water goes.
| Document | How the water is applied | Time per spot | What the helper watches |
|---|---|---|---|
| GM bulletin 18-NA-362, Cadillac ATS and CTS | "Water hose without the nozzle attached" on the upper windshield and roof ditch | Flood "for up to five minutes" | The interior A-pillar |
| GM bulletin 13185, 2013 Cadillac ATS | Hose without nozzle, run under the lifted lace moulding | 1 to 2 minutes | The headliner edge at the top of the A-pillar, by hand |
| Nissan quality action PM988, 2021 Kicks | "Full hose pressure water stream" on the upper corner | At least 2 minutes | The upper corner of the seal, pillar trim lowered |
| GM procedure for the 2011 Saab 9-4X | Pipe test stands, aimed 30 degrees down and 45 degrees rearward at the corner | 155 kPa (22 psi) at the nozzle for at least 4 minutes | Any leak, from inside |
| Daimler Trucks Service Solution 3495 | Hose outlet a few inches away, water poured on | At least 1 minute, so cavities can fill | The dash and pillars |
| Volkswagen tech tip 61-21-01TT, 2011 to 2022 models | Garden hose or sprinkler, "sprayed from the top down to simulate rainfall" | Not stated | Not stated |
| Land Rover Discovery 3 workshop manual | Fine spray, from "the lower end" and "gradually moving up", never straight at the rubber seal | Not stated | Not stated |
Apart from the Saab rig, this is a garden hose, so the test you can run in a driveway is close to the one a dealer runs. Keep the Daimler minimum in mind: water entering high can take a full minute to fill a hidden cavity before one drop reaches the dash.
Top down or bottom up? Both, in that order
Two instructions circulate, and both come from carmakers. Volkswagen sprays from the top down to simulate rain. Land Rover starts at the bottom and works up, the order used in the home hose test for leaks and wind noise. They answer different questions.
- Top down confirms the leak. Water falling like rain over roof and glass together reproduces the complaint. If the car stays dry under that, a hose aimed at odd angles is testing something the car never meets.
- Bottom up locates it. Water that enters high runs down inside the pillar and behind the dash, so a footwell drip after wetting the top edge could have come from anywhere. Wet the lowest zone first, then the next one up, and each new drip can only belong to the zone you just added.
Run the rain test once to prove the leak exists, let the car dry, then go zone by zone from the cowl upward. Volkswagen adds a rule for both: do not spray upward or at unusual angles into the seals.
Why a pressure washer proves nothing
GM's leak test preparation page says sealing materials are designed for "the natural elements" and "cannot take into consideration any artificial conditions, i.e., high pressure car washes", and that its test "has been correlated to the natural elements". Volkswagen says "high pressure wands should not be used"; Daimler and Nissan rule out pressure washers and pressure sprayers. Volkswagen's tip on the 2007 to 2013 Eos hardtop even puts a figure on it: the seals cannot withstand washes generating "up to 3500 PSI". A leak that only appears under a wand is not the leak you are chasing. How washes treat new glass is covered in car washes and your windshield.
The soap and air test that pins the exact gap
A hose tells you the zone. Soap and air tell you the inch: raise the pressure inside the cabin, wet the outside seam with soapy water, and watch for bubbles.
- Seal the cabin. Ford's body diagnosis section tapes off the body vents, closes windows and doors, sets the blower to high and turns recirculation off. Daimler uses fresh air mode, fan on high.
- Mix the solution. Daimler specifies at least a quarter cup of soap per gallon of water, in a spray bottle.
- Spray the seam, not the glass, along the glass-to-moulding and moulding-to-body joints. Ford also names the A-pillar at the fender and the hood hinge area.
- Mark every bubble, then rinse off the soap and turn the blower off.
Dealers sharpen it with shop air. An FCA case file for water at the A-pillar has the technician brush soapy water on the upper corner outside and aim "a rubber safety tip air nozzle with regulated shop air" at it from inside. FCA bulletin 23-031-19 for the 2019 Jeep Wrangler works the lower center moulding gap the same way while someone watches the dash top, a short build window covered in the Jeep Wrangler windshield guide.
A bubble is not always a leak. Daimler warns that the soap test "may also identify areas that will not leak water, even though those areas produce bubbles", and small bubbles outside a suspected area may need no repair. Find the zone with the hose first, then trust bubbles inside it.
When the leak is not the windshield at all
Several bulletins begin as a windshield complaint and end at a body seam.
| Where the water shows | Source found | Document |
|---|---|---|
| Running down the inside of the A-pillar | Pinholes or cracks in the roof ditch laser braze, or voids in the upper flange seam of the opening | GM 18-NA-362, 2014 to 2019 Cadillac ATS and CTS |
| "From around the windshield" | A gap between the roof panel and the roof cap panel | GM 24-NA-129, 2023 and 2024 BrightDrop Zevo 600 |
| At the A-pillars or windshield header | Excessive factory gaps in the body seams; seams sealed, new header seal | Ford 21-2403, 2021 Bronco |
| Dripping onto the dash from above | The roof cap seam above the A-pillars, often mistaken for a marker light | Daimler Service Solution 3498 |
| Wet front carpet, sloshing noise | Missing body sealer at the cowl ends, body joints and wheel well | Nissan NTB14-105, 2014 Rogue |
| Water in the front passenger cabin | A gap in the urethane seal between body and glass | Hyundai campaign 9A6, 2024 Tucson and Santa Cruz |
Only the last row is the glass bond itself, on a build window covered in the Hyundai Tucson windshield page. The Nissan procedure runs water down inside the cowl and checks under the floor spacers before blaming anything, which is why the cowl panel and its seal belong on the list after a replacement; Rogue specifics are in the Nissan Rogue windshield page. A sloshing sound is a clue in two documents: Daimler hears it above the windshield when cab cavities fill, Nissan while driving. A headliner drip on a car with a glass roof belongs with sunroof glass and its drains before anyone cuts urethane.
What the fix should be, by where the gap is
The repair follows the location, and the bulletins use four.
- Seam sealer on a body joint. GM works clear seam sealer into a braze pinhole, Ford lays strip caulk along the Bronco's seams, Nissan adds body sealer wherever none is visible. No glass comes out.
- Sealant into the moulding gap. The Jeep bulletin puts silicone glass sealer between glass and frame, then re-tests once it has dried. Hyundai fills the gap in the urethane seal with a syringe, allowing 0.3 labor hours and recommending shade or indoors.
- Urethane over the existing bead. Ford's 2014 Mustang "glass reseal" lays new urethane over the old, top and sides from inside and bottom from outside, with "all gaps" smoothed into one continuous bead. It "must cure for a minimum of one hour before testing for air or water leaks."
- Glass out. Nissan replaces the Kicks windshield because urethane at the upper corner "may not adhere to the body primer". Ford's body chart answers "gaps in sealant bead" and glass "misaligned or not installed correctly" with new glass. GM removes and reseals the ATS and CTS glass if sealing the braze did not stop the water.
The last group is the glass shop's. A bond that never stuck points to primer and pinch weld prep, a bead with a gap to the single continuous pass described in how a replacement should be done and how windshield urethane holds the glass, and corrosion found under the old bead to what happens when the shop finds rust. Every procedure ends the same way: re-test after the sealer has cured.
Running the test yourself
You need a hose without a nozzle, a helper with a flashlight, a towel and a spray bottle of soapy water.
- Dry the car and clear the cowl, including under the mats, so a new drip is a new drip and leaves are not doing the leaking.
- Prove the leak with falling water over the roof and glass, like rain, while the helper watches.
- Find the zone from the bottom up. Dry off, then flood the cowl and lower edge, each lower corner, each side, and the top edge last: at least a minute per zone, up to five on the upper corners.
- Find the gap with soap in the zone that leaked, blower on high, recirculation off, doors shut.
- Write down the date and zone. That record is what makes a warranty claim stick.
A professional inspection is short. The carmaker labor allowances here run from 0.2 hours (Nissan, and the Jeep inspection) through 0.4 hours (GM's water test with no further action) to 0.5 hours (Daimler's inspection and leak testing), a fair yardstick for any diagnostic fee.
Where this stops applying
- Bulletins are narrow. Each covers named models, often a build window of days, such as Wranglers built July 1 to 5, 2019. They show how leaks are tested and where they have been found, not what is wrong with your car.
- Gasket-set glass is different. For Freightliner trucks, replacing a windshield gasket "unnecessarily may create new leak issues", and a cab pressure test comes first. See semi truck windshields, and for gasket-set glass on older cars, classic car glass.
- A brand new install is not ready for a hose test. Give the urethane the time in the first 48 hours after a new windshield first.
- Glass that comes out needs the full job again. Nissan's Kicks procedure keeps the car parked "for at least 24 hours or until the urethane adhesive has completely cured" and aims the camera afterward, the steps in safe drive away time and when calibration is required.
What to do next
If the glass was replaced recently, take your test notes back to the installer; what a shop owes you is in the replacement warranty guide, and if it will not engage, the route is in disputes and complaints with a glass shop. Pricing a reset elsewhere starts with the cost estimator, and the rest of the job is in the replacement section.
Frequently asked questions
How do I find where my windshield is leaking?
Dry the car, then have a helper sit inside while you run a garden hose with no nozzle over one zone at a time, starting at the cowl and lower edge and working up to the top. Give each spot at least a minute. When water appears, run the blower on high with the doors shut and spray soapy water on the outside seam of that zone: bubbles show the gap.
Can I use a pressure washer to test for a windshield leak?
No. Carmaker procedures rule it out. GM says its sealing materials are designed for natural rain, not high pressure car washes, and Volkswagen, Nissan and Daimler all say not to use pressure wands or sprayers. High pressure can force water past a seal that rain never would, so it can produce a leak that is not the one you are trying to find.
Why is water dripping from my headliner near the windshield?
It can be the upper corners of the windshield bond, but several carmaker bulletins traced this symptom to something else: pinholes in a roof seam, a gap between the roof panel and roof cap, or factory gaps in body seams near the windshield header. A sunroof drain can do it too. A hose test on the roof seam first, with the windshield moulding lifted, separates them.
Can a windshield leak be fixed without replacing the glass?
Often, if the gap is small and in a seam or the moulding channel. Carmaker fixes include seam sealer on body joints, silicone or syringe sealant in the moulding gap, and Ford's reseal that lays new urethane over the old bead. If the urethane never bonded, the bead has gaps, or the glass sits wrong, the procedures call for removing or replacing the glass.
How long should a leak test take at a shop?
Not long. The carmaker labor allowances for a windshield leak inspection in the bulletins reviewed run from 0.2 hours at Nissan and FCA to 0.4 hours at GM and 0.5 hours for Daimler's inspection and leak test. Removing trim to watch the pillar adds time, and the fix is billed separately, but finding the leak itself is a short job.