MTBF / MTTF – The Great Divider
There’s perhaps no other metric in reliability engineering that divides practitioners more than Mean Time Between Failures (MTBF) and Mean Time To Failure (MTTF). While there is a technical difference—MTBF applies to repairable items and MTTF to non-repairable ones—let’s keep it simple and use MTBF to refer to both for the sake of readability.
In the reliability community, two camps have emerged:
Those who see value in the MTBF metric.
Those who see no value, especially when it comes to predicting real-world operational performance or evaluating historical asset performance.
What Is MTBF, Really?
Let’s take a step back and look at what MTBF actually tells us.
Despite common misconceptions:
MTBF is not a life expectancy.
It’s not the average age at which a unit will fail.
It does not tell you how long something will last.
What it does represent is this:
If the MTBF is 20 years, you have a 37% chance that a given unit will last that long without failure.
In other words, out of 100 identical units with a 20-year MTBF, only 37 would still be operating at the 20-year mark. The other 63 have already failed—but when did they fail? That’s the problem MTBF can’t answer.
The failures could occur early or late, clustered or scattered. MTBF treats them as random failures, and the value is simply:
Total operating time ÷ Number of failures
That’s it.
So, What Does That Number Actually Tell You?
Here’s the reality check:
Can MTBF help you improve equipment life? No.
Can it help determine optimal maintenance intervals? Not really.
Can it predict how failure probability changes with age? Definitely not.
MTBF gives you a pinpoint metric, but no shape to the failure curve. It doesn’t inform preventative maintenance strategies. It doesn't account for wear-out mechanisms. It doesn’t support effective asset renewal planning.
This disconnect between design-stage MTBF assessments and actual field performance likely explains why so many reliability projections fall flat in practice.
Why Are We Still Using It?
MTBF is an old-school metric, dating back to the 1960s, long before modern Computerised Maintenance Management Systems (CMMS). Back then, reliability engineers didn’t have access to the rich, real-time performance data we have today.
Sometimes, when failure data is sparse or unavailable, MTBF may be the only option. But when you do have good CMMS data, it’s worth exploring more informative alternatives:
Weibull analysis
Hazard rates
Survival probability curves
Condition-based analytics
These approaches provide real insight into failure modes, degradation trends, and optimal interventions.
And trust me—your asset management and maintenance teams will thank you.
So will your finance director.