The short version
- The rock test finds bad bearings but misses worn ones. No play does not mean no problem.
- Rock at twelve and six for the bearing. Rock at three and nine for steering parts.
- The spin test catches roughness long before play shows up, and it costs nothing.
- Heat is the most underrated check. Compare corners, never judge one in isolation.
You have a noise, you have ruled out the tires, and you want to know whether the bearing is actually bad before you commit to a part.
How to check a wheel bearing comes down to five tests worth doing. None is conclusive alone, which is what a shop diagnosis fee buys you, which is the part most guides skip, so this covers what each one finds and what each one misses.
Get the car up safely first
Every check below needs the wheel off the ground, and several involve pulling hard on a wheel.
Jack stands under a rated point, chocks on the wheels still down, and the transmission in park with the parking brake set. Never rely on a jack alone for anything where you are applying force.
On a car with an electronic parking brake acting on the rear, be aware it may hold the rear wheels and mask play there.
The rock test, and the thing nobody tells you
This is the standard check and it is genuinely useful. It is also the source of more false negatives than any other test in this subject.
Grip the tire at twelve and six o’clock. Push in at the top and pull at the bottom, then reverse it. You are looking for movement, or a knock as the slack takes up.
Then move your hands to three and nine and repeat.
| Where the play is | What it usually means |
|---|---|
| Twelve and six | Wheel bearing, or an upper or lower ball joint |
| Three and nine | Tie rod end, or steering rack play |
| Both directions equally | Bearing, since it allows movement in every plane |
| Stops when a helper holds the brake | Not the bearing. Holding the rotor removes bearing play from the test. |
A wheel bearing has to be substantially worn before it produces play you can feel by hand. Noise usually arrives long before movement does. So a bearing that passes the rock test can still be the source of your growl, and plenty of people conclude the bearing is fine on exactly this evidence.
Treat a positive result as conclusive and a negative result as inconclusive. That asymmetry is the whole point of the test.
The spin test finds what the rock test misses
This is the one that catches worn bearings before they get loose, and it takes about thirty seconds a corner.
Spin the wheel by hand, reasonably fast, and listen. Then do it again with a hand resting on the coil spring or the strut body.
A healthy bearing spins smoothly and almost silently. A worn one produces a rumble or a gritty roughness you can feel through the spring as much as hear.
Compare against the opposite corner immediately. Judging one bearing in isolation is much harder than noticing that one side feels different from the other.
Light pad contact makes a swishing noise that masks bearing roughness, and it is present on most cars. Spin the wheel and listen for a scrape that repeats once per revolution, which is the pad. Bearing roughness is continuous and does not sync with the wheel.
A mechanic’s stethoscope on the hub while a helper spins the wheel sharpens this considerably. A long screwdriver with the handle against your ear does the same job less comfortably.
The heat check
Underrated, and the closest thing to a direct measurement you can take without tools.
Drive far enough to get things warm, ideally without heavy braking near the end, then feel each hub center. Not the rotor, which will be hot regardless.
A failing bearing generates friction heat, so it runs warmer than its opposite number. What you are looking for is a difference between sides, not an absolute temperature.
Warm on both sides is normal. Noticeably hotter on one side is a strong signal. Too hot to keep your hand near, while the other side is merely warm, means the bearing is close to failing and the car should not be driven far.
An infrared thermometer removes the guesswork and removes the risk of burning yourself. Brake drag confuses this test badly, so if one rotor is also much hotter, suspect a sticking caliper before the bearing.
Measuring it properly
If you want a number rather than an impression, a dial indicator gives you one.
Mount the indicator so the plunger tip rests on the machined face of the hub at twelve o’clock, parallel to the axis of the spindle. Grip the wheel at three and nine.
Pull outward while rocking the wheel roughly 45 degrees. Keep pulling, stop rocking, and read the dial.
| Setup | Acceptable end play |
|---|---|
| Sealed hub unit, most modern vehicles | None. Any measurable movement is a fail. |
| Adjustable tapered set, rear-drive front | Up to about 0.010 inch on some applications |
| Where a spec exists | Typically 0.001 to 0.005 inch |
The spec matters. Adjustable bearings that are set too tight fail from heat, and too loose they knock and wear the races. Both errors are common and both are avoidable with a gauge.
What shops use when nobody can agree
A chassis ear settles arguments. It is a set of clip-on microphones with a channel selector, and you drive the car while switching between corners.
The loudest channel is the problem. It bypasses the whole issue of sound traveling through the body and arriving from the wrong direction, which is why rear bearings are so hard to place by ear.
If you are paying a shop a diagnostic charge on a noise nobody can pin down, this is usually what you are paying for, and it is generally money well spent compared to replacing parts speculatively.
The road tests, and their limits
Two things are worth doing before the car goes on stands.
Coast in neutral at the speed where the noise is clearest. If it continues unchanged while the engine drops to idle, it rotates with the wheels. That eliminates the engine and most of the driveline.
Load the corners with gentle swerves on a quiet road. Weight moves outward in a turn, so the bearing that gets louder is on the outside of that turn.
The load test is right more often than not, and it is also wrong often enough that it should never be your only evidence. A badly worn bearing is loud in both directions, cupped tires respond the same way, and a second failing bearing hides behind the louder one. There is more on the noises that imitate a bearing.
How to check a wheel bearing, all five tests together
- Road test. Coast in neutral, then load both directions. Note which corner you suspect.
- Rock test, both planes. Play confirms it. No play proves nothing.
- Spin test with a hand on the spring. This is where worn bearings actually show up.
- Compare against the opposite side. Every check is more reliable as a comparison.
- Heat check after a drive. A clear difference between sides is a strong signal.
- Measure with a dial indicator if you want a number, or you are setting an adjustable bearing.
Two positives from different tests is a confident diagnosis. One positive and four inconclusive results usually means keep listening, particularly if the noise is still early. What the sound itself tells you about the stage fills in the rest of the picture, and the symptom overview covers the signals that are not noise at all.
The side that gets loaded is the one people get backwards, and decoding left from right covers the inversion in detail.
Why a clean rock test proves so little is on the no play page, and separating a worn joint from a worn bearing is on the ball joint page.
Common questions
How to check a wheel bearing without removing the wheel?
Jack the corner clear and rock the tire at twelve and six, then spin it and listen. Both work with the wheel on, and the wheel actually gives you more leverage to detect play.
Should there be any play in a wheel bearing?
On a sealed hub unit, none that you can detect. Some adjustable tapered setups specify a small amount, typically thousandths of an inch, measured rather than felt.
My bearing has no play but it is noisy. Is it bad?
Quite possibly. Noise generally arrives well before play, so a noisy bearing with no movement is a normal stage of failure rather than a contradiction.
How to check a wheel bearing by spinning it?
Yes, and it is the most sensitive simple test available. Listen for rumble and feel for roughness through the spring, then compare against the other side.
How hot should a wheel hub get?
Warm is normal after a drive. What matters is the comparison between sides. Significantly hotter on one corner, with no brake drag to explain it, points at the bearing.
Does holding the brake really change the rock test?
Yes, and it is a useful trick. Clamping the rotor removes bearing play from the equation, so play that disappears when a helper presses the pedal is coming from somewhere else.