---
title: "Symptoms of a Bad Thermostat"
description: "A thermostat fails in two opposite directions and only one of them overheats. We read 501 owner complaints filed with NHTSA to see how each one actually reaches a driver, and why most are not filed as cooling faults."
url: "https://baronauto1.com/car-care/symptoms-of-a-bad-thermostat/"
type: "article"
published: "2026-09-07"
modified: "2026-09-07"
site: "Baron Auto"
disclaimer: "This site is under new ownership and is not affiliated with Baron Auto Emporium dealership."
---

# Symptoms of a Bad Thermostat

> A thermostat fails in two opposite directions and only one of them overheats. We read 501 owner complaints filed with NHTSA to see how each one actually reaches a driver, and why most are not filed as cooling faults.

*Detailing & Maintenance · 25 min read · 5,691 words*

## The short version

- A thermostat is the only part of a cooling system that can fail in two opposite directions. Stuck shut, the engine overheats. Stuck open, it never gets warm — no overheating at all, but a cold heater, a gauge that sits low, poor economy and usually a fault code.
- We read 165,795 owner complaints filed with the federal regulator and pulled out the 501 that name a thermostat. Of those 501, only 65 were filed under the component label a reader would think to search: engine and engine cooling. The most common label was plain engine, on 326 of them.
- 209 of the 501 use overheating language and nothing from the cold side. 37 use cold-side language — no heat, a gauge reading below normal, the code P0128 — and nothing from the hot side. Almost half describe neither, because they name a thermostat as a part someone already replaced.
- The most striking single finding: 56 of the 501 mention a steering warning, against 3 in a control set of 4,156 cooling complaints that never say the word. On several vehicles this failure reaches the driver as reduced power steering and an air conditioning cut-out, with nothing on the dashboard about temperature at all.
- The two directions have opposite calendars. Cold-side reports cluster in winter — 43.2 per cent fall in December, January or February, and 24.3 per cent in December alone. Hot-side reports peak in July, like cooling complaints generally.
- 67 of the 501 name the thermostat housing rather than the valve. That is a leak, not a stuck valve, and it is a different problem with a different urgency.
- The single most useful thing you can do with a suspected thermostat is not replace it. It is establish which direction the fault points, because the two directions send you to completely different places.

Search for the symptoms of a bad thermostat and you get the same list everywhere: overheating, temperature gauge fluctuating, coolant leaks, poor heater performance. The list is not wrong. It is just assembled from what a thermostat is capable of doing rather than from what happens to people, and the difference turns out to matter, because the way this failure actually reaches a driver is frequently nothing like the way the lists describe it.

So we went and counted. NHTSA publishes every complaint an owner files about a vehicle, including the owner’s own words, and that archive is free to query. We pulled it across 30 model lines from 17 makes over the 2010 to 2020 model years, scanned 165,795 complaints, and kept the 501 whose narrative uses the word thermostat. Then we kept a second set as a control: 4,156 complaints filed under the engine and engine cooling label whose narratives never mention one — the rest of the cooling system, described by the same kind of person on the same form.

What comes out of that is a more awkward and more useful picture than the symptom lists give. This page sets out what the 501 say, what the counting can and cannot support, and what to do with a car that is showing one of these signs. It is not a repair guide, and there is no procedure here for getting under a car. It is about recognising which of two opposite failures you are looking at, and deciding what happens next.

## What 501 owner reports actually say

Start with the size and shape of the sample, because that decides what any of it is worth. 274 requests went out, one per make, model and model year. 269 answered; 5 never did and are recorded as failures rather than quietly dropped. Between them those requests returned 165,795 complaints, of which 4,221 — 2.5 per cent — carry the engine and engine cooling component label at all.

501 narratives use the word thermostat. That is 0.3 per cent of the archive, which sounds small until you remember it is a single valve in a single subsystem of a car with several thousand parts. Those 501 come from 14 makes and 27 of the 30 model lines queried. Three model lines produced no thermostat mention whatsoever, and that absence is worth stating plainly rather than glossing: this is not a fault that shows up everywhere.

Then the finding that reorganised this whole page. Of the 501, only 65 carry the engine and engine cooling label — 13 per cent. Whatever a thermostat is in engineering terms, it is not filed as a cooling problem in the federal record. The label it most often carries is plain engine, on 326 of them, 65.1 per cent, against 24.5 per cent for the archive as a whole. Electrical system carries 89. Power train carries 81. Unknown or other carries 80. Steering carries 42.

Read down that list and you can see a diagnostic problem taking shape. The cooling label is genuinely enriched here — 13 per cent against 2.5 per cent across the archive is a fivefold concentration, so the classifiers are not blind. But 87 per cent of the time the complaint went somewhere else, because the owner arrived at the intake describing the thing they experienced, and the thing they experienced was rarely a temperature problem.

**Figure: The label is not the part**

A two-column figure of the NHTSA component labels carried by the 501 complaints whose narrative names a thermostat, from ENGINE on 326 of them down to STEERING on 42, with the two labels the article measures against the whole archive shown against it.

The cooling label is genuinely enriched — 13 per cent against 2.5 per cent across the archive is a fivefold concentration, so the classifiers are not blind — and it is still a minority. 87 per cent of the time the complaint went somewhere else, because the owner arrived at the intake describing the thing they experienced, and the thing they experienced was rarely a temperature problem. Two cautions. A complaint can carry several component labels at once, so each narrative is counted once per distinct label it holds and the figures do not add to a hundred. And a label is assigned during intake rather than by an engineer after a teardown: it records what the complaint was understood to be about, which is exactly the point being made here, and it is not evidence about what broke. These are complaints owners chose to file, not failure rates, and no make is compared with another.

One caution about the figure before it gets over-read. A complaint can carry several component labels at once, so each narrative is counted once per distinct label it holds and the shares do not add to a hundred. And a label is assigned during intake, not by an engineer after a teardown. It records what the complaint was understood to be about, which is exactly the point being made here, but it is not evidence about what broke.

## One part, two opposite failures

A thermostat is a temperature-operated valve between the engine and the radiator. Cold engine, valve shut: coolant circulates inside the engine only, so it warms up quickly. Up to temperature, valve opens: coolant reaches the radiator and heat leaves the car. It is a wax pellet and a spring doing a job that sounds trivial and is not, because an engine that runs cold and an engine that runs hot are both damaged engines, in different ways and on different timescales.

That gives this part a property nothing else in the cooling system has. A water pump has one failure direction. A radiator has one. A thermostat has two, and they point in opposite directions.

**Stuck shut** is the one everyone knows. Coolant never reaches the radiator, so the temperature climbs, and it climbs fast — often within a few miles of a cold start, on a car that was fine yesterday. The speed and the timing are the tells. It does not care whether you are in traffic or on a motorway, because the fault is upstream of everything airflow does.

**Stuck open** is the one that gets missed for months. The valve never closes, coolant circulates through the radiator constantly, and the engine never reaches the temperature it was designed to run at. Nothing overheats. Nothing steams. The heater blows lukewarm or cold, the gauge sits below where it usually sits, fuel economy drops, the engine runs richer than it should, and eventually the engine management module notices and sets a code. Almost nobody experiences that as a cooling fault, and the narratives show it.

Classifying the 501 on vocabulary — the words the owner used, not a diagnosis anyone confirmed — gives 209 narratives that use overheating language and no cold-side language, 37 that use cold-side language and no overheating language, and 9 that use both. That leaves 246, 49.1 per cent, using neither.

The 246 are not a failure of the classifier; they are a finding. Read a sample of them and the pattern is consistent: the thermostat appears in a list of parts that have already been replaced. Spark plugs, coils, water pump, thermostat, sensor. It is named as a thing that was tried, in a complaint about something that is still happening. That is the strongest signal in this dataset about how the part is actually treated in the field — it is a cheap early guess, and it is often not the answer.

## The failure that arrives as a steering warning

Here is the number that made this page worth publishing. Of the 501 thermostat narratives, 56 — 11.2 per cent — mention a steering warning: reduced steering assist, a power steering fault, steering described as disabled. In the control set of 4,156 cooling complaints that never say the word, that language appears 3 times, 0.1 per cent.

Air conditioning shows the same shape. 71 of the thermostat narratives mention it, 14.2 per cent, against 87 of the far larger control set, 2.1 per cent. And the diagnostic code P0128 — the code for engine coolant not reaching its regulating temperature, which is the stuck-open signature — appears 27 times among the 501 and not once in the control.

The mechanism behind all three is the same and it is worth understanding, because it explains an entire class of complaint that looks like it has nothing to do with cooling. Modern cars manage electrical load against engine temperature. When the module decides the engine temperature reading is wrong or out of range, it starts shedding load and disabling functions that depend on knowing how hot the engine is: the compressor stops, electrically assisted steering drops to a reduced-assist mode, and a message appears telling the driver about steering or air conditioning. The temperature gauge, meanwhile, may be sitting somewhere unremarkable, or reading low, or reading nothing.

So the driver’s experience is a steering warning and no air conditioning. Not one word about coolant. That is how the complaint gets filed under steering or electrical system, and it is how a reader arrives at a page about thermostats several weeks and one wasted diagnosis later.

**Why we compare these two sets carefully and not casually.** The two populations are not written the same way. NHTSA’s file mixes complaints owners typed themselves with complaints a call handler wrote up from a phone call, and the handled ones follow a template that begins with the words “The contact owns a” and then the vehicle. Those are 104 of the 501 thermostat narratives, 20.8 per cent, and 3,446 of the 4,156 control narratives, 82.9 per cent — because a call handler classifying a cooling complaint reaches for the cooling label, while an owner typing free text does not. Templated prose is shorter and more formulaic, which is enough to move any wording comparison a few points. It is not remotely enough to explain 56 against 3, or 27 against zero. Where the gap is small, we have not drawn a conclusion from it.

The check engine light is the case in point. It appears in 30.7 per cent of thermostat narratives and 40.4 per cent of the control — a difference well inside what the writing-style gap could produce on its own, so nothing is claimed from it. What can be said is the plain version: 154 of these owners mention a light, which makes an amber lamp one of the most common ways this failure is first noticed, and our guide to [what the check engine light actually means](https://baronauto1.com/car-care/what-does-check-engine-light-mean/) covers the trap that follows, which is that a code names a circuit rather than a component. P0128 does not say the thermostat is faulty. It says the coolant is not getting hot enough, which a stuck-open thermostat causes and a handful of other things also cause.

## Half of this failure is only visible in winter

Cooling complaints have a season. Sort the 4,156 control narratives by the month the incident happened and July is the peak at 10.1 per cent, with December the trough at 6.4 per cent. Summer holds 29.1 per cent of the year’s reports and winter 22.5 per cent. That is what you would expect from a system that is being asked to reject heat.

Split the thermostat narratives by direction and the two halves come apart completely.

The 209 hot-side narratives behave like the control, only more so: 30.1 per cent in summer, 24.9 per cent in winter, peaking in July at 12.9 per cent. The 37 cold-side narratives do the opposite. 43.2 per cent of them fall in December, January or February. 24.3 per cent fall in December alone. Only 10.8 per cent fall across the whole of June, July and August, and August contributes none at all.

**Figure: Two directions, two calendars**

A grid comparing the incident-month profile of the 37 cold-side thermostat narratives with the 209 hot-side ones and with a control of 4,156 cooling complaints that never name a thermostat, by peak month and by the share falling in winter and in summer.

Winter is December, January and February here, and summer is June, July and August. The comparison itself is sound in a way the wording comparisons on this page are not, because both halves are drawn from the same 501 narratives — nothing about how the complaint was written differs between them, only the season. But 37 narratives can carry a shape and not a measurement: the December share of 24.3 per cent rests on 9 complaints, and August contributes none at all. Treat the winter clustering as a direction. The explanation is almost embarrassingly simple — a thermostat stuck open produces no symptom a driver can feel in warm weather, so the fault is present in July and invisible in July, and it becomes detectable on the first genuinely cold morning. Which is the argument for running the heater to full hot on a viewing even in August. Shares within each group, from complaints owners chose to file; not failure rates.

The explanation is almost embarrassingly simple, and it is the single most practical thing on this page. A thermostat stuck open produces no symptom that a driver can feel in warm weather. The engine runs cooler than it should, which does not announce itself; the heater is not being used, so nobody notices it is not working; economy drifts by an amount nobody tracks. The fault is present in July and invisible in July. Then the first genuinely cold morning arrives, the driver asks for heat, and there is none.

Two consequences follow. The first is for anyone buying a car in summer: a stuck-open thermostat is one of the few significant faults that a warm-weather test drive is structurally incapable of finding, which is a reason to run the heater to full hot on a viewing even in August and watch the gauge come up. The second is for anyone whose heater has just gone cold in December — you are not necessarily looking at a fault that started in December. You are looking at one that became detectable in December.

**What 37 narratives can and cannot carry.** The cold-side group is small, and a December share of 24.3 per cent rests on 9 complaints. That is enough to show a shape and not enough to put a number on the size of the effect, so treat the winter clustering as a direction rather than a measurement. The comparison itself is sound in a way the wording comparisons above are not: both halves are drawn from the same 501 narratives, so nothing about how the complaint was written differs between them. Only the season does.

## What it gets mistaken for

The reason a thermostat is worth its own page is that both of its failure directions have a well-established impostor, and in each case the impostor is the cheaper or more obvious guess. This is what the narratives keep describing, and it is where money gets spent in the wrong order.

| What you actually notice | What it usually gets blamed on first | What separates them |
| --- | --- | --- |
| Temperature climbs hard within a few miles of a cold start, at any speed, in any traffic | Low coolant, or a failed cooling fan | Indifference to speed. A fan fault only bites when the car stops moving, and a coolant loss shows a falling level and usually a coloured crust somewhere. A valve that never opens does not care what the car is doing |
| Gauge sits below its usual position, heater lukewarm or cold on a fully warmed engine | A blocked heater core, a stuck blend door, or low coolant | Where the heat is missing. A heater-side fault leaves the engine at normal temperature and the cabin cold. A valve stuck open leaves the whole engine cold, so the gauge is low as well as the cabin |
| Check engine light, code P0128, no temperature complaint at all | The coolant temperature sensor, because the code names it | The code describes a condition, not a part. A sensor reading a genuine low temperature and a sensor reading wrongly look identical on a scan tool and are told apart by measurement, not by the code |
| Steering assist or air conditioning messages on a car that drives normally | A steering fault, a compressor, or an electrical gremlin | Whether the temperature reading is plausible. These are load-shedding responses to a temperature the module does not trust; the steering and the compressor are usually fine |
| Slow coolant loss, crust or damp around the housing, level dropping over weeks | The thermostat, because the housing carries its name | Leak against valve. A cracked plastic housing is a coolant loss with a normal-working valve inside it. The symptoms are the symptoms of losing coolant, not of a stuck valve |
| Overheats only at sustained speed or on a long climb, system completely full, no noise | The thermostat, as the standard cheap first move | The load pattern. A valve stuck partly shut can do this, but so can an eroded water pump impeller or a radiator blocked internally — and those two make no noise and leak nothing |
| Temperature climbs, drops sharply, climbs again | A faulty gauge or sender | A valve opening late and stuttering does this, and so does air moving through the system in pockets after a coolant loss. Both are real; a gauge that is simply lying is the least likely of the three |

The row that costs the most money is the last-but-one. A car that overheats only under sustained load, with a full system and nothing audible, is the classic case where a thermostat is replaced because it is the cheapest part in the argument, and the fault continues because the actual limit was somewhere else. Our guide to [what to do when a car overheats](https://baronauto1.com/car-care/car-overheating/) works through that whole decision from the other end, sorting the fault by when it appears rather than by what the part is.

## The housing is not the valve

67 of the 501 narratives, 13.4 per cent, name the thermostat housing specifically. In the control set the phrase appears zero times, which is unsurprising, and the number is high enough among the 501 to deserve separating out.

On a great many modern engines the thermostat sits inside a moulded plastic housing that also carries hose connections and often the temperature sensor. Plastic in that position goes through several thousand heat cycles and eventually goes brittle, and the failure is a crack and a leak rather than a valve that sticks. The owner reports it as a thermostat problem because that is what the part is called and that is what the invoice says.

It matters because the two failures behave differently and want different responses. A cracked housing is a coolant loss: a slowly falling level, a coloured crust on the outside of the housing or below it, a top-up habit that becomes routine. The valve inside it may be working perfectly. The danger is not the valve but the loss, because a system that is quietly going low will eventually be found out by a hot day and a hill. A stuck valve, by contrast, produces a temperature problem with the coolant level exactly where it should be.

Distinguishing them is mostly a matter of looking rather than testing. Coolant dries to a coloured crust rather than staying wet, and that crust marks the actual source instead of the lowest point of the drip, so a housing that has been weeping usually shows where. A stuck valve leaves nothing to see at all.

## What else was in the same paragraph

One more cut through the 501, and it is the one that says most about how this fault behaves over time. Counting what else each narrative names: 80 mention an engine replaced, blown or seized. 79 mention a water pump. 64 mention a radiator. 63 mention a coolant leak or a low level. 38 mention a head gasket.

Two of those deserve reading against the control before anybody panics. Engine replacement language appears in 16 per cent of thermostat narratives and 36.3 per cent of the control — the thermostat group is, if anything, the less catastrophic of the two. Towing shows the same direction: 12.6 per cent against 25.1 per cent. Whatever else a thermostat complaint is, it is not usually the worst thing that happens to a cooling system.

The water pump and radiator counts are the other way round: 15.8 per cent against 1.7 per cent, and 12.8 per cent against 1.5 per cent. Part of that is topical — a narrative already listing cooling parts is likely to list several — but the pattern in the text is consistent and specific. These parts get replaced together, sometimes because a workshop is doing the sensible thing while the system is open, and sometimes because each one was tried in turn.

The head gasket count is the one to hold carefully. 38 narratives name one, 7.6 per cent, against 4.1 per cent in the control. Nothing in this data establishes an order of events, and a head gasket is far more often the consequence of prolonged overheating than the cause of it. What the number supports is a warning, not a diagnosis: if you are being told a thermostat explains a car that has been overheating repeatedly for months, the question worth asking is what the repeated overheating has already done.

The archive also carries an odometer reading in the small subset of complaints written to the call-intake template, which states the failure mileage in a fixed sentence. That gives 92 usable readings among the thermostat narratives, with a median of 80,000 miles, against 3,075 readings and a median of 95,000 for the control. Both numbers are right-censored in the same way — these cars are still being driven, so the readings that would sit above the median have not happened yet — and 92 readings out of 501 narratives is a thin base. The honest reading is a hint that thermostat trouble arrives somewhat earlier than the cooling system’s other failures, not a mileage to plan around.

## Deciding what to do next

The practical structure of this fault is unusual, and it is worth stating in the form of a decision rather than a list of symptoms.

**If the temperature is climbing, the priority is not the diagnosis.** Damage from overheating accumulates by the minute. Heater to full hot with the blower up, air conditioning off, ease the throttle, stop somewhere safe, and leave the cap alone until the engine is genuinely cold — a pressurised system releases its contents upward at whoever opens it, and it is the classic serious injury in this whole subject area. That sequence is worth reading before it is needed rather than after.

**If nothing is climbing, you have time, and you should use it.** A stuck-open thermostat is not an emergency. It is a slow tax: worse economy, a richer mixture, more wear during a warm-up phase that never ends, a cabin that does not heat and windows that do not clear properly, which is the part that turns it into a safety question in winter. None of that requires a decision this afternoon. All of it justifies establishing the cause properly rather than buying a part.

**Establish direction before parts.** The single most useful question is whether the engine is running hot or running cold, and the answer is nearly free. On a cold start, watch how the gauge behaves as the engine warms: a normal system rises steadily to its usual position and stays there. A system whose valve never opens rises past it. A system whose valve never closes struggles to reach it, and a heater that stays cool after a good run says the same thing from a different direction. That observation costs nothing and it sends the diagnosis to two entirely different places.

**Then get the code read rather than guessed.** If a light is on, the code is data. P0128 points at the cold side. A code naming the coolant temperature sensor is set by a thermostat far more often than by the sensor, which is exactly why a code should be treated as a starting point rather than a parts order.

**Why there is no figure here.** A price printed today keeps looking authoritative long after it has stopped being right, and nothing on the page tells a later reader how much time has passed. Quotes for the same job also vary by region, by vehicle and by what else has to come apart to reach the part.

What survives is the list of things that decide the number. That is what this page sets out, so you can work out where a specific quote sits rather than measuring it against an average that was never about your car.

What drives the number here is not the part, which is among the cheapest things in an engine bay. It is access. On some engines the thermostat sits in an obvious housing at the end of the top hose and the job is short. On others it is under an intake manifold, behind a timing cover, or integrated into a housing assembly that is only sold complete with sensors and hose stubs attached. On engines where reaching it means disturbing the timing drive, the sensible workshop does the pump and the belt at the same time, and the conversation stops being about a thermostat at all. Establish which of those your car is before you take a quote as high or low, because the same part number can sit behind wildly different amounts of work.

## If you are looking at a car to buy

Two checks, both free, both quick, and one of them is season-dependent in a way almost nobody accounts for.

First, insist on a cold start. A car that is already warm when you arrive has had its most informative ten minutes taken away from you, and that is true of this fault more than most: the entire warm-up curve is the evidence. Watch the gauge come up from cold and settle. Too far, too fast, or never quite arriving are three different findings.

Second, run the heater to full hot regardless of the weather. In August that feels absurd and it is the only way to detect a stuck-open thermostat before the first cold morning, which the seasonal split above shows is when everyone else finds out. If the engine is at normal temperature and the vents are not hot, you have found something, whether it is the thermostat, the heater circuit or a coolant level nobody has checked in years.

Both of those belong inside a proper pre-purchase inspection rather than instead of one, and neither requires anything more than standing beside the car and paying attention for ten minutes. And because cooling faults do appear in recall campaigns, and the remedy is free at a franchised dealer regardless of how many owners a car has had, it is worth [checking the VIN for open recalls](https://baronauto1.com/vehicle-history/how-to-check-a-car-for-open-recalls/) before paying anyone to diagnose something a manufacturer has already conceded. 151 of the 501 narratives mention a recall or a technical bulletin, which tells you how often owners go looking for one.

## Common questions

### What are the symptoms of a bad thermostat?

They divide into two opposite groups. Stuck shut: the temperature climbs quickly after a cold start, at any speed and in any traffic, often with a warning message or steam. Stuck open: the engine never reaches normal temperature, the heater blows lukewarm or cold, the gauge sits low, fuel economy drops and a fault code appears — commonly P0128, which describes coolant that is not reaching its regulating temperature. In the 501 owner complaints we read, 209 used overheating language and nothing from the cold side, and 37 used cold-side language and nothing from the hot side. A third group, 246 of them, named a thermostat only as a part that had already been replaced.

### Can a bad thermostat cause a check engine light with no overheating?

Yes, and it is one of the most common ways the stuck-open failure is first noticed. The engine management module monitors how long the coolant takes to reach its regulating temperature and sets a code when it never does. That code appeared 27 times in our thermostat narratives and not once in the control set of cooling complaints. Nothing overheats, nothing steams, and the only outward sign may be an amber lamp and a heater that is not as hot as it used to be.

### Why does my car show a steering or air conditioning warning when the problem is cooling?

Because modern cars shed electrical load when the module does not trust the engine temperature it is being given. The compressor stops and electrically assisted steering drops into a reduced-assist mode, and those are the things the dashboard tells the driver about. In our sample, 56 of 501 thermostat narratives mentioned a steering warning, against 3 in 4,156 cooling complaints that never named a thermostat. The steering and the compressor are usually fine; the temperature signal is what is wrong.

### Is it safe to drive with a thermostat stuck open?

It is not an emergency in the way overheating is, but it is not nothing either. An engine that never reaches operating temperature runs richer, wears faster and uses more fuel, and in cold weather the real risk is visibility: a heater that cannot clear a windscreen is a safety problem before it is a comfort one. Get it established and fixed on a normal timetable rather than an urgent one.

### Is it safe to drive with a thermostat stuck shut?

No. That is straightforward overheating, and damage accumulates by the minute rather than by the mile, so a short journey at a crawl can be as harmful as a long one. Stop, let it cool properly, and get it recovered rather than nursed home.

### How do I tell a thermostat from a water pump or a radiator?

Mostly by when the trouble appears. A thermostat stuck shut sends the temperature up fast and from cold, and it does not care about speed. A pump failing at its impeller and a radiator blocked internally both cope at low load and fail under sustained load, so they produce a car that is fine in town and climbs on a long run with a completely full system. A pump failing at its bearing or seal announces itself with noise or a weep. Our page on [what drives a water pump replacement](https://baronauto1.com/car-care/water-pump-replacement-cost/) covers why the drive arrangement changes that job so much.

### My mechanic replaced the thermostat and the problem came back. What now?

That pattern is well represented in the complaints we read: almost half the narratives name a thermostat inside a list of parts already tried, in a report about a fault that is still happening. The part is cheap and it is a reasonable first move, but it is a guess unless something confirmed it. The next step is measurement rather than another part — a pressure test on a cold system, a check of what the temperature sensor is actually reporting against reality, and a combustion leak test if coolant is disappearing with no external leak.

### Is a leaking thermostat housing the same thing as a bad thermostat?

No, and the distinction is worth insisting on. 67 of our 501 narratives named the housing rather than the valve. A cracked plastic housing is a coolant leak with a working valve inside it, and its symptoms are the symptoms of losing coolant: a level that keeps dropping, a crust or damp patch around the housing, a top-up habit. A stuck valve produces a temperature problem with the level exactly where it should be.

### Does a thermostat fail at a particular mileage?

Nothing in this data supports a number to plan around. The complaints that state a failure mileage in the intake template — 92 of them among the thermostat narratives — have a median of 80,000 miles, against 95,000 for the wider cooling set, and both figures are right-censored because the cars are still on the road. Treat it as a hint that thermostat trouble arrives a little earlier than the cooling system’s other failures, and not as a service interval.

### Which cars have the most thermostat complaints?

We are deliberately not publishing that as a ranking. The count for any model reflects how many were sold, how long they have been on the road and how inclined their owners are to file, at least as much as it reflects how often a part fails, and our vehicle list is 30 model lines chosen for breadth rather than a census. What the sample does support is the shape of the failure, which is the same on every make in it.

## Sources and further reading

- [NHTSA Office of Defects Investigation complaint database](https://www.nhtsa.gov/nhtsa-datasets-and-apis)
- [NHTSA recall lookup](https://www.nhtsa.gov/recalls)
- [40 CFR 86.1806-17 — onboard diagnostics](https://www.ecfr.gov/current/title-40/chapter-I/subchapter-C/part-86/subpart-S/section-86.1806-17)

Recall, complaint and safety-rating figures on this page were retrieved from the federal databases above on August 19, 2026. Federal data changes — re-check any VIN before you rely on it.

---

*This site is under new ownership and is not affiliated with Baron Auto Emporium dealership.*

Canonical source: https://baronauto1.com/car-care/symptoms-of-a-bad-thermostat/