How to Calculate MTBF (Mean Time Between Failures)

MTBF, or Mean Time Between Failures, is the average operating time between one failure and the next. The formula is MTBF = Total Operating Time ÷ Number of Failures. A higher MTBF means a more reliable, repairable asset. It is one of the core reliability metrics in any maintenance program.

The MTBF formula

MTBF = Total Operating Time ÷ Number of Failures.

The two inputs sound simple but need care. Total Operating Time is the time the asset was actually running and available to fail, not the full calendar and not time it sat switched off. Number of Failures counts only genuine breakdowns, not planned stops, changeovers, or scheduled maintenance. Get those two definitions right and the calculation is straightforward.

Calculate your MTBF

Enter total operating hours and the number of failures to get MTBF and the implied failure rate. The defaults match the worked example that follows.

A worked example

Take a packaging line tracked over a year: 4,000 operating hours and 8 failures.

Metric	Formula	Calculation	Result
MTBF	Operating time ÷ failures	4,000 ÷ 8	500 h
Failure rate	1 ÷ MTBF	1 ÷ 500	0.002 /h

On average the line runs 500 hours between failures. Fix the most common failure mode and cut it to 6 failures, and MTBF rises to 667 hours, a 33 percent reliability gain from removing two breakdowns. That is the kind of before-and-after MTBF is built to show.

MTBF vs MTTF

MTBF and MTTF look similar and are often confused, but they apply to different things.

• MTBF (Mean Time Between Failures) is for repairable assets. You fix the pump and it runs again, so there is time between failures.
• MTTF (Mean Time To Failure) is for non-repairable items. A bearing, a fuse, or a bulb is replaced rather than repaired, so you measure time to the single failure.

Use MTBF for systems you maintain and return to service. Use MTTF for components you simply swap out. See MTTR vs MTBF for how reliability and repair speed work together.

MTBF and failure rate

Under the common assumption of a constant failure rate, MTBF is just the inverse of that rate.

Failure rate (λ) = 1 ÷ MTBF. An MTBF of 500 hours implies a failure rate of 0.002 failures per hour. This is useful for combining components and for spares planning, but it rests on the assumption that the failure rate is steady, which is not always true.

Converting MTBF to failure rate and FIT

MTBF is often expressed in two other forms, especially for electronic components: failure rate and FIT. They describe the same thing from different angles, and the conversions are simple, as long as you assume a constant failure rate, which is the flat middle of the bathtub curve.

Failure rate, written as lambda, is the inverse of MTBF:

Failure rate = 1 / MTBF

So an asset with an MTBF of 10,000 hours has a failure rate of 0.0001 failures per hour.

FIT, which stands for failures in time, expresses that rate as the number of failures per billion hours of operation. It is the standard unit for component reliability:

FIT = 10^9 / MTBF, and equivalently MTBF = 10^9 / FIT

That same 10,000-hour MTBF works out to 100,000 FIT (a billion divided by 10,000). A more reliable part with an MTBF of one million hours is 1,000 FIT. The smaller the FIT number, the more reliable the component. The one caveat: these conversions hold only while the failure rate is roughly constant, so they describe the useful-life period, not early-life or wear-out failures.

Calculating reliability from MTBF

MTBF also answers a more useful question than "how long on average": what is the probability that an asset runs for a specific length of time without failing? That probability is reliability, written R(t), and for a constant failure rate it follows the exponential formula:

R(t) = e^(-t / MTBF)

where t is the time period you care about and e is the base of the natural logarithm, about 2.718.

Take an asset with an MTBF of 10,000 hours. The probability it runs a 1,000-hour mission without failing is e^(-1,000 / 10,000), which is e^(-0.1), or about 90.5 percent. Over 5,000 hours it falls to e^(-0.5), about 61 percent.

One result surprises people: at t equal to the MTBF itself, reliability is not 50 percent. It is e^(-1), about 36.8 percent. An asset is more likely than not to have failed by the time it reaches its mean time between failures. That is a property of the exponential distribution, and it is why MTBF on its own, without the reliability curve behind it, can paint a falsely reassuring picture of how long equipment will actually last.

How MTBF feeds availability

MTBF rarely travels alone. Paired with MTTR, the average time to repair, it gives you inherent availability, the share of time an asset is ready to run.

Inherent Availability = MTBF ÷ (MTBF + MTTR).

If the line above runs 500 hours between failures and takes on average 4 hours to repair, its inherent availability is 500 ÷ 504, or 99.2 percent. Raising MTBF and lowering MTTR both push that number up, from opposite directions: one means failing less often, the other means recovering faster.

Availability is also the first of the three factors in OEE, which is why reliability work shows up directly in your overall effectiveness score, not just in a maintenance report.

What counts as a good MTBF

There is no universal good number. A good MTBF depends entirely on the asset, its duty, and your industry, so a conveyor motor and a turbine are not comparable. Two things matter more than the absolute figure.

• The trend. Rising MTBF means improving reliability. That is the signal you actually manage to.
• The bathtub curve. Failure rates are usually high early (infant mortality from install or defects), low and steady through useful life, then high again at wear-out. MTBF describes the flat middle, so a single number can be misleading near the ends of an asset's life.

How to improve MTBF

MTBF improves when you remove failures, not when you fix them faster. Repairing faster lowers MTTR; it does not change how often the asset breaks. So the work is about eliminating failure modes, usually starting with the one that happens most often.

In the worked example, removing the two most common breakdowns lifted MTBF by a third. That is the leverage point: a small number of recurring failure modes tends to account for most of the downtime, so fixing the top one or two moves the number more than chasing every rare event. The practical loop is to rank failures by frequency, fix the root cause of the biggest one, and watch the trend rather than any single reading.

Common mistakes

• Treating MTBF as a guaranteed lifespan. An MTBF of 500 hours does not mean the asset will last 500 hours. It is a statistical average across a population and a period, not a countdown timer.
• Counting planned downtime as failures. Scheduled maintenance and changeovers are not failures. Including them inflates the failure count and understates true reliability.
• Assuming a constant failure rate during wear-out. Late in life the simple formula misleads. Watch the trend, not just the latest value.
• Calculating from too small a sample. A handful of failures gives a noisy number. The more operating hours and failures behind it, the more meaningful the result.

From calculating MTBF to improving reliability

Calculating MTBF tells you how reliable an asset is and whether it is trending the right way. What it does not tell you is which failure mode to attack first, what is driving the breakdowns, or whether the fix is worth the spend on a given line. Those answers are buried in work orders, sensor histories, and production data.

This is the part SteelTree handles. It connects to the systems already holding that data, and where the history is thin or scattered, it captures what it needs directly from the work itself. It turns the trend into a decision: which assets are losing reliability, what is driving it, and the highest-return action to take next, with the reasoning attached. And because it captures the reasoning behind each decision, the system compounds. It gets sharper at your plant the longer you run it.

It is how SteelTree drives equipment reliability from the same numbers.

See how SteelTree can transform your operational processes →