What causes thrust bearing failure? There isn't one answer.
When a maintenance team calls and asks "what causes thrust bearing failure," they usually want a single answer. Lubrication. Overload. Bad batch. Give me the one thing, and I'll fix it.
I've handled NSK bearing orders for roughly eight years. I've personally submitted wrong part numbers, approved orders that came back rejected, and watched customers replace the same thrust bearing three times before someone asked a basic question. Between rework and expedited shipping, my mistakes cost our team somewhere around $4,800. That's not counting the credibility hit.
Here's what I've learned the hard way: thrust bearings fail four different ways, and each one needs a different fix. If you can't tell which scenario you're in, you'll end up repeating the same mistake—just with a new part number.
Let's break it down by timeline. Because when it failed is usually the fastest way to figure out why.
Scenario A: It failed within days of install
If a thrust bearing gives out within 48 to 72 hours of installation, the bearing itself is almost never the problem.
The most common error: installing it backwards. Thrust ball bearings have a shaft washer and a housing washer. They look similar, they aren't interchangeable. Put them in the wrong way and the shaft washer can't contact the rolling elements properly. It skids. Then it wears. Then it fails.
The second most common error: alignment. Thrust bearings are extremely sensitive to angular misalignment between the shaft and the housing seat. A few taps with a hammer—a slight tilt that seems harmless—can cut service life from years to weeks.
We had a batch of 40 thrust bearings go into a conveyor line back in 2019. All 40 failed within a week. The customer blamed the product. Turns out they'd aligned the mounting flanges to each other instead of to the shaft. About 0.3 degrees of misalignment. For a thrust ball bearing, that's fatal.
What I learned from that one: when you're working with a standard thrust bearing, check the mounting angle. It is more sensitive than you think. If you're building a spindle running under preload and handling combined loads—that's a different conversation, and you should be looking at angular contact bearings. But a standard thrust ball bearing needs a flat seat and proper alignment. Full stop.
Scenario B: It failed after a few weeks
Six to eight weeks is the lubrication window. That's when the film starts breaking down, or contamination starts grinding away at the raceways.
Most thrust bearings are open—they rely on the surrounding environment to keep the grease in and the dirt out. In a foundry, a dusty workshop, or a machining center with chips flying, open bearings don't stand a chance if they aren't sealed properly.
I had a customer in 2021 who went through two batches in four months. The bearing itself was fine. The problem was they were running an open thrust ball bearing in an environment with metal fines in the air. No seal. No shielding. The particles embedded in the raceway, scored it, then flaked.
Here's where my experience contradicts the usual advice. Everyone says "check lubrication first"—and that's true. But the bearing industry has published failure studies for decades, and roughly 40-50% of bearing failures trace back to lubrication and contamination issues. Most people assume it's an overload problem. More often than not, it's dirt.
What to do: if contamination is your problem, switch to a sealed variant. Or better—redesign the housing to add a labyrinth seal. For harsh environments, don't keep replacing the open bearing. Upgrade to a sealed thrust bearing or a fully enclosed unit. That's where the money actually goes.
Scenario C: It failed again. And again.
If you're on your third thrust bearing in the same application, the bearing isn't the issue.
I made this mistake early on. We replaced a thrust bearing three times in about 14 months. Same failure mode each time—spalling on the raceway at around the 4-6 month mark. I kept thinking it was a material batch issue. It wasn't. The alignment between the shaft and the housing was off just enough to load the bearing way beyond what it was designed for.
But here's the thing most people miss: check the roller chain sprocket. If you've got a roller chain sprocket running in a chain drive and it's slightly out of line, it pushes axial load onto the shaft. That load has to go somewhere—usually straight into your thrust bearing. I've seen a team replace two thrust bearings before someone finally measured the sprocket. It was off by 1.5mm. That tiny offset was pushing constantly against the bearing.
Your checklist: check roller chain sprocket alignment. Check for axial load coming from somewhere else on the shaft. Re-verify your equivalent dynamic load calculation. If there's any dynamic axial component in your application, standard thrust ball bearings have a fairly low design load capacity—you might need a cylindrical roller thrust bearing instead.
Scenario D: You chose the wrong bearing type entirely
This is the one that gets overlooked—and the one that cost us the most money.
Sometimes a thrust bearing failure isn't a quality problem. It's a design problem. If you're running a spindle that needs axial and radial load support under preload, a thrust ball bearing is the wrong tool. You need angular contact bearings—paired, with the right preload. I've been through the NSK angular contact bearings catalog more times than I can count helping customers match pairs. The pairing tolerance class (ABEC 5 / P5 level precision matters at speed) and the contact angle will determine your service life more than any other specification.
If space is the constraint—small motors, compact robot joints, medical devices—a standard thrust bearing is probably too big. NSK miniature ball bearings go down to 1mm bore. But look carefully at the load capacity: miniature bearings can't handle big axial loads. I've had to tell customers that. It's not what they want to hear, but it saves them a return six months later.
And if you're just doing general rotation and don't need to handle thrust—deep groove ball bearings are usually the answer. They handle axial load in both directions, they cover a wide speed range, and they're cost-effective. I'd honestly say over 60% of the orders we process end up being deep groove ball bearings. Not because the others don't sell—but because most applications don't need anything more complex.
Real talk: not every application fits the NSK catalog. If you're running extreme loads at high speed with contaminated cooling, talk to an application engineer first. Don't just click "buy" on the first thrust bearing you find. That's how you end up with a $3,200 order that fails in six weeks.
How to figure out which scenario you're in
Check the timeline. That's the simplest filter.
- Failed within days → Installation. Check the mounting angle. Don't trial-and-error it.
- Failed within weeks → Lubrication or contamination. Open it up. Look for grinding marks vs. spalling.
- Failed within months → System problem. Check alignment, check the roller chain sprocket, check whether the bearing type even fits.
- Failed after expected life → Normal wear. Re-evaluate bearing size. Maybe you just need a bigger one.
And before you replace the part, ask one question: "Is this bearing the right one for the job?" If the system hasn't changed and you're on your third replacement, the answer is usually no.
We put together a 12-point pre-check list in January 2024 after one too many of these situations. It's not fancy—just a one-page checklist that covers alignment, sprocket condition, load calculation, and bearing type selection. In the first 18 months we killed three orders that would have failed. One of them was a $3,200 thrust bearing assembly that was headed into a spindle application. The customer wasn't happy when I said "we shouldn't sell you this." But I'd rather lose the order now than explain the failure in six months.
That's the trade-off. If I can't tell you honestly when not to buy something, you shouldn't trust me when I tell you to buy something else.

Leave a technical comment