Mean time between maintenance (MTBM): how often you have to touch the machine at all

A reliability engineer is reviewing a critical conveyor. Its mean time between failures looks healthy, so on paper it is running well. Then she tallies up every maintenance action that actually stopped the conveyor last year, not just the breakdowns, but the six preventive visits and the three predictive interventions too, and the picture shifts: the machine is being interrupted, on average, every fifty hours.

That broader view is mean time between maintenance. Where MTBF counts only the failures, MTBM counts every time you have to stop the asset and lay hands on it.

What it actually measures

MTBM is the average running time between consecutive maintenance actions that interrupt an asset's function, of any kind: corrective, preventive or predictive. Because it counts them all, it reflects the total maintenance burden an asset places on you, not just how often it fails.

How to work it out

MTBM = Operating time (hours) / Number of maintenance actions

An asset runs 1,000 hours and is interrupted by 10 corrective, 6 preventive and 3 predictive tasks.

MTBM = 1,000 / (10 + 6 + 3) = 1,000 / 19 = 52.6 hours

Reviewing the schedule, the team finds two preventive tasks that the monitoring data shows are simply not needed, and drops them. The next 1,000 hours bring 17 interruptions instead of 19, lifting MTBM to 58.8 hours. A small change, but a measurable one, and the asset now spends more of its life doing useful work between interruptions.

What it tells you that MTBF cannot

The reason MTBM earns its own place is that it drives the availability you actually live with. Reliability engineers describe availability in tiers. The inherent figure, the designer's best case, is built from how often the asset fails and how long it takes to repair, mean time between failures and mean time to repair. But the operational availability you experience on the floor is built instead from MTBM and mean downtime: how often you have to touch the asset at all, against how long it is truly down each time. Where MTBF asks only about breakdowns, MTBM captures the full rhythm of interruption, including all the planned stops, which is why it tracks the real, day-to-day availability of the machine more honestly than failures alone ever could.

What good looks like

As with the other mean metrics, there is no universal benchmark; the right figure depends on the asset and what it does. What you want is the trend moving upward over time, a sign that your maintenance strategy is growing more effective and the asset is being disturbed less often to keep it running well.

Where it can mislead you

• Count only the actions that genuinely interrupt the asset. Work done online, without stopping it, does not belong in the tally.
• A rising MTBM is only good news if you know why it is rising. Earned by cutting needless corrective work, it is real progress; achieved by quietly deferring necessary preventive tasks, it is risk piling up out of sight. The two look identical on the chart, so always check the mix of corrective and preventive behind the number.
• Use it alongside MTBF, never instead of it. MTBM shows the whole maintenance burden; MTBF isolates the failures. You need both lenses to see clearly.
• A low MTBM on a critical asset is a direct invitation to dig deeper, with a root cause analysis or a hard look at the strategy that keeps demanding so much attention.

MTBF asks how often a machine fails; MTBM asks how often you have to touch it at all, for any reason, and it is the version that feeds the availability your operators actually feel. Raising it the right way means your strategy is doing more with less interference, which is the quiet hallmark of a plant that has its maintenance genuinely under control.