How a Windshield Crack Actually Grows
The short answer
- A crack extends when tension at its tip beats what the glass there can hold, so triggers matter far more than elapsed time.
- Glass cracks in two modes: slow growth of a fraction of an inch over weeks as moisture attacks the strained tip, and fast fracture that crosses the panel before you finish hearing the ping.
- The visible line stops short of the real tip. View the crack under raking light before you measure it, because shops routinely measure longer than drivers do.
- A crack that stopped has not healed. The tip stays sharp and the threshold for the next run drops slightly every month it sits open.
- Cracks steer to stay perpendicular to the greatest tension, so a hook toward the perimeter means the bonded edge zone is loaded and the damage belongs in the replacement conversation.
- Repair works by removing the stress concentrator: resin bonds the fracture faces so the tip is no longer a free surface, which stops growth without making the mark disappear.
A crack grows for one reason: the tension pulling at its tip is greater than what the glass right at that tip can hold. Everything else, the cold morning, the pothole, the car wash, is just a way of adding tension. Understanding what happens in the fraction of an inch ahead of the visible crack tells you why the same damage can sit still for two years and then cross your windshield in one second, and why nobody, including a good technician, can give you a date.
The crack is longer than the line you can see
Start with the thing most drivers get wrong. The dark line on your glass is not the crack. It is the part of the crack that has opened wide enough to scatter light. Ahead of it, the fracture usually continues as a hairline that is far too fine to catch your eye at normal viewing angles, and ahead of that is the tip itself, a separation between two atomic planes.
You can often see the hidden extension. Park so that low sun or a flashlight beam skims the surface at a shallow angle, get your eye down near the glass, and look along the line rather than at it. The faint continuation that appears at raking light is real crack, and it is why a shop measures a crack longer than you did over the phone. The technique is part of measuring damage the way a technician does.
This matters for two decisions. It moves your length measurement closer to the repair limit, and it means a resin repair has to reach further than the visible line to seal the actual tip.
Why a sharp tip multiplies the load
Glass is brittle in the technical sense: it has no way to relieve a concentrated load. A steel bracket under too much stress at a sharp corner will yield slightly, blunting the corner and spreading the load around. Glass cannot do that. Whatever tension the panel is carrying gets funneled into the tip and multiplied there, and the sharper the tip, the larger the multiplier.
That single fact explains most of what shops tell you about break shapes:
- A bullseye is the stable break. Its boundary is a smooth cone with a rounded edge. There is no sharp tip anywhere on it, so the panel's stress flows around it much the way water flows around a rock.
- A star break is the unstable one. Each leg ends in a tip that is as sharp as fracture can make it. A star break of the same overall diameter as a bullseye carries several loaded tips instead of none, which is why the urgency tables on this site rank it higher regardless of size.
- A crack is a single tip with a long lever. The longer the crack, the more panel area is feeding stress into that one point, which is why long cracks are progressively easier to extend and short ones are harder to start.
- Edge damage sits in the worst place. The bonded perimeter is where the panel carries load into the body, so a tip there gets the highest background tension the windshield has to offer. That is the mechanism behind the two inch rule described in edge cracks.
Break shapes and what each one is called are cataloged in windshield damage types, and the damage identifier will walk you through naming yours.
Cracks grow at two completely different speeds
This is the part that reconciles the contradictory stories drivers hear. Glass fracture has a slow mode and a fast mode, and they look nothing like each other.
In the slow mode, the tension at the tip is below the level needed to snap bonds outright, but water in the air reaches the strained bonds at the tip and reacts with them, breaking them one at a time. The crack creeps. It can advance a hair over weeks with no event at all, and it advances faster in humid air than in dry air. This is the reason a crack under steady tension is never truly static.
In the fast mode, the tip crosses the threshold and the crack accelerates. Once it is running, it releases stored energy faster than it consumes it, so it keeps going until it runs out of driving force. It is over before you finish reacting to the sound.
| Regime | What is happening at the tip | What you observe | What pushes it to the next regime |
|---|---|---|---|
| Below threshold, dry | Tension at the tip is low and there is little moisture to attack the strained bonds | Nothing, for months or years | Any rise in panel tension: a gradient, a flex, a load |
| Slow growth | Moisture is breaking strained bonds at the tip a few at a time | The line seems slightly longer than you remember, with no event to blame | Humidity, sustained tension, and time |
| Fast fracture | Driving force exceeds the glass's resistance and the tip accelerates | A ping, and a line that is inches or feet longer than it was a second ago | Already there. It stops on its own or not at all |
| Arrest | The tip has run into lower tension or the panel's stored energy is spent | The crack ends mid-panel, often with a slight hook | Nothing. The tip is parked, still sharp |
| Re-initiation | A parked tip is loaded past threshold again | The same crack extends weeks later, often on a cold morning | A new thermal gradient, flex event or pressure pulse |
Two practical conclusions follow. A crack that appears to grow with no cause is not a mystery, it is slow growth. And a crack that has been quiet all summer has not become safe, it has been parked.
The threshold moves, which is why timing feels random
It would be convenient if each crack had a fixed load it could survive. It does not, because both sides of the comparison move. The driving side is the sum of every tension acting on that tip at that moment: the residual stress locked into a bonded panel, the gradient from sun on one half of the glass, the twist from a driveway apron taken at an angle, the pressure pulse from a door slammed with the windows up, the wind load at highway speed. Those stack. None of them alone is usually enough, which is why cracks so often run during an ordinary combination of ordinary things.
The resisting side moves too. It falls as the tip is chemically nibbled by moisture, and it varies with the microscopic condition of a tip you cannot inspect. That is why two identical looking star breaks on two identical cars behave differently, and why any confident prediction about your specific glass is guesswork. To rank your own risk factors rather than guess a date, use the scoring framework in will my windshield crack spread.
Quiet is not the same as healed. Glass has no repair mechanism of its own. A tip that stopped moving in March is exactly as sharp in September, sitting in the same panel, waiting for the same combination of loads. Elapsed time without growth is not evidence of stability, it is evidence that the threshold has not been reached yet.
Why one crack runs straight and another wanders
A running crack steers. It turns continuously to keep itself perpendicular to the direction of greatest tension, because that is the path that releases the most energy. So the shape of a crack is a map of the stress field it ran through, and you can read it backward.
| Path shape | What the stress field was doing | Typical origin |
|---|---|---|
| Straight run across open glass | Uniform tension over a large area, direction constant | Impact break releasing into the middle of the panel |
| Long gentle arc | Tension direction rotating slowly across the panel, typical of a thermal gradient | Defroster or sun heating one region while another stays cold |
| Hook or J shape near the perimeter | The restrained bonded edge bending the field sharply | Stress cracks and cracks that wander into the edge zone |
| Fork or branch | Far more driving energy than the crack needed, so it split rather than accelerate further | A hard impact, or a long crack running under high load |
| A step or jog partway along | The field changed abruptly at that point | A previous repair, the frit boundary, an embedded antenna or a bonded bracket |
The most useful reading is the hook. A crack that curves as it approaches the perimeter and then runs parallel to it is telling you the edge zone is loaded, and cracks that arrive from an edge with no impact anywhere along them are the classic signature described in stress cracks versus impact cracks.
Why a running crack stops
A crack in flight is spending stored elastic energy. It stops when the accounting turns against it, which happens in a handful of recognizable ways.
- The load was transient. A thermal gradient is a temporary condition. Ten minutes into the drive the panel has evened out, the tension that started the crack is gone, and the tip parks wherever it was.
- It ran out of the tense region. Tension is not uniform across a windshield. A crack heading from a loaded edge zone toward the calmer middle of the panel is running downhill into lower and lower driving force.
- It ran into compression. Where a region of glass is being pushed rather than pulled, a crack has nothing to open it. This is why cracks so often stall as they approach a warm zone on a cold day.
- It spent its energy on branching. Splitting into two tips costs energy, and a forked crack often dies shortly after the fork.
- The interlayer absorbed the rest. The plastic layer between the two glass plies is tough and stretchy, so fracture energy dumped into it is not available to extend the crack. This is also why the outer ply can break while the inner one stays perfect, covered in what a windshield is made of.
Which of these applies to your crack, and how much comfort to take from it, is the subject of the companion page on why some cracks stop growing.
Why it starts again three weeks later
An arrested crack keeps two things: a sharp tip and a permanent home in a stressed panel. Slow growth then works on that tip continuously, so the threshold that has to be reached for the next run is a little lower every month. Meanwhile the loads come around on a seasonal cycle, and the first hard freeze, the first heat wave and the first long highway drive after a season of city commuting are all step changes in the tension the panel sees.
That combination produces the pattern shops see constantly: a crack appears, runs a few inches, sits untouched through mild weather, then extends across the glass on the first cold morning of the year. It is not the cold that has been sitting there waiting for the crack, it is the crack that has been sitting there waiting for the cold. The specific mechanism of that morning is in cold weather and defroster cracks.
What the physics tells you to do
Everything above points at one action: eliminate the tip. That is what a resin repair actually does. It fills the void, bonds the two fracture faces together and gives the tip a filled, bonded boundary instead of a free surface where stress can concentrate, which is why a properly executed repair restores strength and stops the growth even though the damage remains faintly visible. The procedure step by step is in how chip repair works, and the same logic scaled up, with its lower success rate, is in long crack repair.
Two next steps depending on where you are. If you are still deciding whether this is repairable at all, run the numbers through the repair or replace tool, since length, edge distance and location decide it. If you already know it is long or edge involved, price the job including any camera work with the cost estimator and read when replacement is required before you book.
Frequently asked questions
How fast does a windshield crack travel when it runs?
Far faster than you can watch. Once a crack passes its threshold it accelerates and keeps going until the driving force runs out, so the whole event is finished before you have finished reacting to the noise. That is why drivers describe a crack as appearing rather than growing, and why nothing you do inside the car in that moment changes the outcome.
Can a crack grow while the car is parked overnight?
Yes, in both modes. Slow growth continues any time the tip is under tension and there is moisture in the air, and a parked car sitting through a temperature swing develops exactly the gradient that pushes a tip past threshold. Cracks that appear or extend overnight with nobody near the car are ordinary, not evidence of vandalism or a defect.
Does humidity really affect how a crack behaves?
It affects the slow mode. Water reaching the strained bonds at a crack tip reacts with them and breaks them one at a time, so a loaded crack advances more readily in damp air than in dry air. The effect is small day to day and easily swamped by a cold start, but it is the reason an untouched crack can be measurably longer a month later.
Why did my crack stop halfway across the windshield?
Usually because the load that started it was temporary or because the tip ran into calmer glass. A thermal gradient evens out within minutes of driving, and tension falls as a crack leaves the bonded perimeter for the middle of the panel. The stop tells you where the stress was that day. It does not tell you the crack is finished.
If a crack has not moved in a year, can I just leave it?
You can, but understand what you are betting on. The tip is unchanged and the panel around it is unchanged, so you are wagering that no future combination of cold start, pothole and highway run reaches the threshold. The cost of losing that bet is a replacement plus any camera calibration instead of a repair that may have been deductible free.