MTBF (Mean Time Between Failures) is the primary metric for measuring equipment reliability. If MTTR tells you how fast you recover from failures, MTBF tells you how often you have to.
What MTBF Means
Mean Time Between Failures is the average operating time between one failure and the next, for a repairable asset. It’s calculated over a population of events (multiple failures of the same asset, or multiple instances of the same asset type) and expressed in hours, days, or another time unit depending on the context.
A higher MTBF means higher reliability — the asset runs longer before failing. A lower MTBF means failures are frequent.
MTBF applies specifically to repairable assets — items that can be fixed and returned to service. For non-repairable items (a light bulb, a fuse, a single-use component), the equivalent metric is MTTF (Mean Time To Failure), which measures how long the item runs before it fails for the first and only time. The math is similar, but the interpretation is different: MTTF describes the lifespan of an item, while MTBF describes the reliability of a repairable system over repeated cycles.
The MTBF Formula
MTBF = Total Uptime / Number of Failures
“Total uptime” is the total time the asset was operational — running and available — over the measurement period, not including time spent in repair.
Worked Example
You’re evaluating the reliability of an air compressor over a 12-month period. Your maintenance records show:
- The compressor was operational for 8,400 hours over the year (accounting for repair downtime and planned shutdowns).
- It experienced 6 unplanned failures during that period.
MTBF = 8,400 / 6 = 1,400 hours
On average, the compressor runs 1,400 hours between failures. With two 8-hour shifts running 5 days a week (roughly 4,160 hours per year), that works out to a failure approximately every 4-5 months.
Whether that’s acceptable depends on how critical the compressor is, what a failure costs, and whether PM can extend the interval.
What Counts as a Failure vs. Planned Maintenance
This distinction matters for accurate MTBF calculation:
Count as a failure:
- Any unplanned stoppage where the asset cannot perform its intended function
- Failures caught during inspection before full breakdown (the asset was found in a failed or near-failed state during a PM check)
Do NOT count as a failure:
- Planned preventive maintenance shutdowns
- Scheduled overhauls
- Shutdowns for operational reasons (production schedule, shift end, planned downtime)
Including planned maintenance events in your failure count will artificially lower your MTBF, making equipment look less reliable than it is. The goal is to measure unplanned failure frequency specifically.
If your records don’t distinguish clearly between planned and unplanned stoppages, your MTBF calculation will be unreliable. This is one practical reason to use structured work orders with a clear work type field rather than a single undifferentiated log.
How to Use MTBF to Set PM Frequency
MTBF and preventive maintenance are closely related. If you know that an asset typically fails after 1,400 hours of operation, you can design a PM schedule to intervene before that point.
A common approach: set PM intervals at 30-50% of your observed MTBF. If your air compressor fails on average at 1,400 hours, a PM interval of 600-700 hours gives you a reasonable margin. The exact fraction depends on:
- The consequence of failure (higher consequence = more conservative interval)
- The cost of each PM event (more expensive PM = push the interval out)
- Whether failure modes are age-related (wear-based failures respond well to interval PM) or random (some failures are not preventable by interval-based PM)
As your PM program runs and you accumulate more data, you can adjust intervals based on actual findings. If technicians are consistently finding nothing to address at the 600-hour PM, the interval may be too conservative. If they’re consistently finding near-failures, it may need to come in.
How to Improve MTBF
A rising MTBF trend on critical assets is one of the clearest signs that a maintenance program is working.
Implement a PM program targeting the right assets. Not all failures are preventable, but many are. A consistent PM schedule that catches lubrication shortfalls, worn belts, and contamination before they cause failures will extend MTBF directly.
Root cause analysis on recurring failures. If the same asset fails repeatedly for different stated reasons, there’s likely an underlying cause that hasn’t been addressed. Repeat failures that look unrelated on the surface — a bearing here, a seal there — can share a root cause like misalignment, vibration, or improper installation.
Operator condition monitoring. Train operators to recognize early warning signs on the equipment they run daily: unusual sounds, vibration, temperature, or fluid condition. Early detection means smaller repairs, less collateral damage, and a failure caught before it becomes a breakdown.
Parts and lubrication quality. Substituting non-specification parts or lubricants to save cost in the short term often shortens equipment life. Track this — if a part change coincides with a drop in MTBF on a class of assets, that’s your signal.
The Limitations of MTBF
MTBF is a useful metric, but it has real limitations worth understanding:
It assumes a constant failure rate. MTBF is derived from reliability theory that assumes failures are random and the asset is equally likely to fail at any point in its life. In practice, many assets fail more at the beginning of their life (infant mortality) and more at the end (wear-out). MTBF doesn’t capture this “bathtub curve” — it gives you an average that may not match any specific period well.
It’s a lagging indicator. MTBF tells you about past performance. It doesn’t predict when the next failure will occur. An asset with an MTBF of 1,400 hours doesn’t become “due” to fail at hour 1,400 — the next failure could be at hour 200 or hour 3,000.
It requires complete failure records. MTBF is only as accurate as your data. If some failures were handled informally and never logged, your failure count is understated and your MTBF is overstated.
MTBF alone doesn’t determine availability. An asset with a high MTBF but a very long MTTR (slow to repair when it does fail) can still have poor availability. Always look at MTBF alongside MTTR to understand the full reliability picture.
For the MTTR side of this relationship, see How to Calculate MTTR. For a concise definition, see the MTBF glossary entry.