The average operating time between one failure and the next for a repairable system. A server, a production line, or a fielded device can be scored this way when each failure is repaired and the equipment returns to service. The result is a reliability figure for the population under study, not a guarantee that any single unit will last that long.
Repair time sits outside this average. Hours spent restoring service are counted on their own, so a fast fix does not make the gaps between failures look longer, and a slow fix does not make them look shorter.
What the Average Counts
Divide total operating time by the number of failures in the same period. The quotient is the average gap from one failure to the next.
- Operating time: Count hours the repairable items were actually running. Time powered off, stored, or held as a spare is not operating time.
- Failures: Count stops that need repair before the system can work again. A planned shutdown for an upgrade is not a failure.
- The next interval: After service is restored, counting starts again. The repair itself is not part of the gap.
What a Single Figure Does Not Promise
The average describes a group of runs, not one unit’s life. One unit can fail early. Another can run well past the average. Publishing the figure as a warranty length overstates what the arithmetic can support.
Conditions matter. The same design can post a different average under a heavier load, a hotter room, or a dirtier power supply. Compare figures only when the duty, the environment, and the failure rules match. There is no universal hour count that every industry should hit.
Compared with Repair Time
How long a repair takes is Mean Time to Repair (MTTR). One measure is the stretch of useful operation. The other is the stretch of downtime after something breaks. Teams need both when they estimate how much of a period the system will be available, because availability depends on the gap between failures and on how quickly service comes back.
| Question | Mean Time Between Failures | Mean Time to Repair |
| What it measures | Average operating time from one failure to the next | Average time to restore service after a failure |
| Which clock runs | Time the system is operating | Time from failure until the repair is done |
| What it describes | Reliability of a repairable system | Speed of recovery |
| What it is not | A promise for any single unit | A substitute for the time between failures |