What an MTBF Number Actually Tells You (And What to Ask For Instead)
Manufacturer MTBF claims run 40,000 to 100,000+ hours, self-reported under a protocol nobody audits. Here is what to actually put in the RFQ instead.
Buyers increasingly write an MTBF requirement into the RFQ, and vendors are happy to hand over a number. Somewhere between 40,000 and 100,000-plus hours, usually. The problem is not that the number is a lie. The problem is that it is self-reported by the manufacturer, measured under a test protocol the manufacturer itself defines, and almost never independently audited. It is a marketing figure with a decimal point of precision attached to it. Useful context, not a fact you can act on by itself.
What do manufacturer MTBF claims actually look like across the market?
Pull MTBF figures from enough vendor datasheets and a pattern shows up fast: the range is wide and the methodology is invisible. Declared numbers commonly run from 40,000 hours at the low end to 100,000 or more at the high end, with a rough industry-wide average cited around 62,000 hours, which works out to roughly seven years of continuous, around-the-clock operation (PatentPC’s downtime and reliability data roundup).
Seven years sounds reassuring on a spec sheet. It should also make you ask a harder question: what exact test produced that number, and does it look anything like your line?
Why isn’t a 60,000-hour claim comparable to an 80,000-hour claim?
Because there is no standardized third-party test protocol that every manufacturer runs against a common baseline. Each vendor defines its own sample size, test environment, duty cycle, and definition of what counts as a failure, and that protocol is almost never published next to the headline hours (LCBOM’s overview of reliability evaluation methods for industrial robots). A 60,000-hour figure and an 80,000-hour figure sitting on two datasheets are not two measurements of the same test. They are two different, undisclosed tests, and the gap between them tells you nothing about which robot is actually more reliable on your floor.
That gap matters most when buyers try to compare reliability across brand tiers, and it is exactly where the research for this piece ran into a wall worth naming. One frequently repeated claim holds that the reliability gap between Chinese and Japanese tier-one robot brands narrowed to under 15% by 2025. Another, equally confident-sounding source puts Chinese robots at roughly 20,000 hours against roughly 80,000 hours for established foreign brands, a 4x gap, not a narrow one. Both numbers get cited like settled facts. Neither traces to a primary, checkable source, and the two cannot both be true. That contradiction is not a footnote. It is the actual evidence for the point of this article: when two authoritative-sounding MTBF comparisons disagree by that much and neither can produce its underlying methodology, the cross-brand MTBF claims circulating online are not something a buyer can safely act on. If you want the country-of-origin quality angle in more depth, with sourcing that holds up, see our full look at Chinese industrial robots. This piece is not trying to relitigate that one, only to flag why a bare MTBF number is the wrong tool for that comparison.
What actually causes robot downtime on the floor?
Here is the part an MTBF number does not cover at all. Most downtime on an industrial robot cell is not the arm itself failing. It is dominated by tooling, integration issues, and peripherals: end-of-arm grippers, vision systems, conveyors, and fixturing, the equipment bolted around the robot rather than the robot’s own base hardware (PatentPC’s robotics maintenance cost and operating efficiency data; practical planning angle from Standard Bots on industrial robot maintenance).
Put plainly: a robot with a stellar MTBF claim can still run inside an unreliable cell if the peripherals around it are weak. Buyers who fixate on the arm’s headline reliability number while treating the gripper, the vision system, and the conveyor interface as an afterthought are optimizing the wrong line item. Maintenance planning that only budgets for the arm, and not for the tooling and integration layer around it, tends to get surprised on the same schedule every time. This is the same story we found building out the seven-year total cost of ownership stack, where maintenance, not the arm price or the electricity bill, turned out to be the layer that actually separates a cheap robot to own from an expensive one.
Why does this matter more now than it used to?
Because the base of robots you have to keep running keeps growing. The IFR’s World Robotics 2025 report puts 2024 annual installations at 542,076 units, the second-highest count in history, and the global operational stock of industrial robots at 4,663,698 units, up 9% year over year (IFR World Robotics 2025 executive summary).
More robots in the field, compounding every year, means more downtime-cost exposure sitting on more RFQs. The due-diligence gap between a headline MTBF number and a verified reliability picture does not shrink as the installed base grows. It compounds right alongside it, on every additional line that goes live trusting a spec-sheet figure it never checked.
Does IRH’s own database say anything useful here?
Only as a proxy, and it is worth being honest about the limit of that. We do not carry an MTBF field for any robot in our database, and we are not going to publish or estimate one, because there is no independently verifiable number to put there. What we can check is something adjacent: how many robots in our catalog have a manufacturer actually publish a datasheet or manual link at all. As of this writing, 295 of 336 robots in the Industrial Robotics Hub database, 88%, carry at least one verified documents link.
That is not a reliability measurement. Publishing a datasheet does not make a robot more durable. But it is a reasonable due-diligence starting signal: a brand willing to publish its own documentation in the open is giving you more to verify against than a brand that will not put anything on paper at all. The 12% of our catalog with no public documents on file is the same category of buyer risk as an MTBF number nobody can audit, just one layer earlier in the process. Long-term parts and software support carries a related risk once you own the robot, which is why controller-generation churn matters for the same reason: see how controller platforms vary by brand for what that looks like when a manufacturer changes its own software stack out from under an installed base.
What to actually put in the RFQ instead of trusting the MTBF line item
None of this means ignore reliability. It means stop treating one self-reported number as the whole answer, and ask for the things a vendor can actually be held to.
| Ask for | Why it matters more than the MTBF headline | Who can verify it |
|---|---|---|
| The MTBF test protocol itself | Sample size, environment, duty cycle, and failure definition tell you whether the number applies to your application at all | The vendor, in writing, before the number is worth citing |
| Warranty term length | A contractual commitment, not a statistical estimate | Legally enforceable, sits in the contract |
| Local field-service response-time SLA | A fast local response often matters more to real uptime than the MTBF figure itself, since most downtime is fixed by a technician, not prevented by hardware | The service contract, checkable against local branch coverage |
| Spare-parts lead-time commitment | A robot waiting on a part is down regardless of how good its MTBF claim was | The parts agreement, checkable against actual delivery history from references |
Every row in that table is something you can hold a vendor to after the sale. An MTBF number on a spec sheet is not. If a supplier will not commit to a warranty term, a service SLA, or a parts lead time in writing, that tells you more about their real confidence in the robot than any hours figure they are willing to print.
The honest version
An MTBF claim is not worthless. It is a starting number from a party with every incentive to make it look good, tested against a method they chose and rarely show you. Treat it the way you would treat any other vendor-supplied projection: ask what is behind it, and then go get the terms that hold up regardless of whether the projection turns out to be right. The used-robot buyers reading our neutral checklist for a used industrial robot already know this instinct. A seller’s claim is a starting point for questions, not a stopping point for due diligence. A new robot’s MTBF line deserves exactly the same treatment.
Frequently asked questions
What does an MTBF number actually mean for an industrial robot? +
MTBF, mean time between failures, is a manufacturer's own estimate of how many operating hours pass, on average, before the robot fails. It is not a guarantee and not an independently audited figure. It is measured under a test protocol the manufacturer itself defines, covering its own sample size, environment, and duty cycle, and that protocol is rarely published alongside the headline number. Treat it as a marketing claim with a number attached, not as a lab result.
Can I trust a manufacturer's MTBF claim? +
You can use it as a rough signal, but you cannot act on it alone. Declared MTBF figures across the market commonly range from 40,000 to 100,000-plus hours, with no standardized third-party test protocol behind those numbers across vendors. A 60,000-hour claim from one brand and an 80,000-hour claim from another are not necessarily measured the same way, so comparing the headline numbers side by side is comparing two different, undisclosed test methods, not two directly comparable robots.
What actually causes industrial robot downtime? +
Most downtime on a robotic cell is not the robot arm itself breaking. It is dominated by tooling, integration work, and peripherals: grippers, vision systems, conveyors, and fixturing. A robot with an excellent MTBF claim can still sit in an unreliable cell if the equipment built around it is weak, which is why MTBF alone answers a narrower question than most buyers assume they are asking.
What should I ask for instead of an MTBF figure in an RFQ? +
Ask for the test protocol behind the MTBF number, not just the headline hours: sample size, test environment, duty cycle, and the definition of failure used. Then ask for the warranty term in writing, the local field-service response-time SLA, and spare-parts lead time commitments. Those four items are independently verifiable and contractually enforceable, which a self-reported MTBF figure is not.
Compare these robots