Maintenance and Reliability KPIs: Benchmarks for Availability, MTBF and MTTR
The reliability KPIs that matter: how to measure each, the levers that move it, and which benchmark figures have a cited public source.
Use these benchmarks as reference points for spotting a gap, not as targets. Your own trend per asset says more about your plant than any single published figure, so let the ranges below tell you where the biggest loss is likely to be.
Every figure shows its basis. A superscript number links to a public source in the Sources list, and the sentence gives that source's definition and scope, because reliability KPIs are defined in more than one way. A figure marked no cited source is kept exactly as originally published because no public source stating it was found, so treat it as illustrative.
At a glance
Maintenance and reliability KPI benchmarks at a glance
Words in parentheses give each figure's original framing or the source's scope. Not stated means the original guide gave no figure for that column. The DOE FEMP figures are facilities maintenance benchmarks taken from NASA in 2000, the NIST figures come from a survey of U.S. discrete manufacturers using 2016 data, and the OEE and availability averages are cross-industry2, 3.
Method: How the figures in this guide are sourced
A figure with a superscript number is stated by the cited source, listed under Sources with its publisher, location in the document and access date, and the sentence gives its definition and scope. A figure marked no cited source is kept exactly as originally published because no public source stating it was found. It is illustrative, not measured industry data. Worked examples show arithmetic only and are not benchmarks.
Equipment availability
Equipment availability is the headline KPI, and the benchmark spread is wide. World-class continuous operations sustain 95 percent or higher2. That is the equipment availability benchmark in the Department of Energy's Federal Energy Management Program (FEMP) O&M guide, taken from NASA in 2000; it is a facilities maintenance benchmark, not one measured on continuous process plants2.
Typical plants run lower, and the number depends on the definition. In a Swedish sample of 94 OEE values from different companies, availability averaged 78.9 percent, with unplanned stops costing 9.6 percent and setup losses 11.5 percent1.
A 2020 NIST survey of U.S. discrete manufacturers, using 2016 data, counted only unplanned downtime. It averaged 7.8 percent of planned production time3; setups and planned stops are not included.
The gap is rarely one big failure; it is dozens of 6 to 15 minute stops no cited source. Track availability weekly per asset and watch the trend line, not the single number.
For example, a line stuck at 85 percent while the plant target is 92 percent is bleeding roughly 7 points, which on a 4,000 hour year is 280 lost production hours worth chasing.
How to measure
Divide the hours each asset is available to run by total hours in the period.
State whether setups and planned stops count as lost time, because published figures differ.
Review the trend per asset every week.
What moves it
Eliminating the short stops that add up, not only the big failures.
Faster setups.
The MTBF, MTTR and PM levers in the sections below.
MTBF (mean time between failures) targets depend on asset class, so benchmark within a family rather than across the plant. The Department of Energy's pumping system sourcebook makes the same point: because systems operate in a broad range of service environments, it is difficult to characterize MTBF for each component7.
A well maintained centrifugal pump should show MTBF above 6,000 running hours no cited source; a poorly maintained one fails under 1,500 no cited source. Conveyors and packaging equipment with many wear points may target 400 to 800 hours, and that is acceptable for the class no cited source.
The lever is defect elimination: precision alignment, correct lubrication intervals and root cause closure on repeat failures. Teams that close the top three recurring failure modes typically lift MTBF 30 to 60 percent within two quarters no cited source.
For pumps, the sourcebook adds that seal and bearing makers publish estimated MTBF, and a time to failure well short of it should trigger a failure-cause assessment7.
How to measure
Divide operating hours by failure count, per asset.
Compare assets only within the same family.
Flag any asset that regresses month over month.
What moves it
Precision alignment.
Correct lubrication intervals.
Root cause closure on the top recurring failure modes.
MTTR (mean time to repair) is the responsiveness KPI, and shorter is almost always cheaper than more reliable. World-class MTTR for line equipment sits under 2 hours no cited source; typical shops run 4 to 8 hours no cited source because parts, information and skills are not staged.
The levers are concrete: kitted spares, single page repair procedures and cross-trained technicians. Staging a critical spare kit alone commonly cuts MTTR by 30 to 40 percent no cited source.
Measure MTTR by asset and by failure mode, because a plant-wide figure can hide the one failure mode that deserves a dedicated fix.
For example, a plant average of 5 hours can hide one failure mode averaging 14 hours.
How to measure
Divide total repair time by the number of repairs, per asset.
Split the result by failure mode.
Decide whether the clock includes waiting for parts and people, and keep that rule fixed.
Planned maintenance percentage separates proactive shops from reactive ones. World-class organizations run 80 to 90 percent of maintenance labor hours as planned and scheduled work4, 3, with unplanned emergency work under 10 percent2.
Those citations come with scope. DOE describes continually top-performing facilities as under 10 percent reactive, 25 to 35 percent preventive and 45 to 55 percent predictive, a program mix rather than labor-hour shares4. In NIST's survey of U.S. discrete manufacturers, the least reactive quarter of plants averaged 14.4 percent reactive maintenance3.
Reactive plants invert this, with 50 percent or more of hours consumed by breakdowns4, 3. DOE, citing a 2000 study, puts more than 55 percent of an average facility's maintenance resources and activities in reactive work4.
U.S. discrete manufacturers averaged 17.3 percent predictive, 31.8 percent preventive and 45.7 percent reactive maintenance in 2016, and the most reactive quarter of plants averaged 81.5 percent reactive3.
The payoff is measurable. Plants that invested more heavily in preventive or predictive maintenance had on average 44 percent less downtime and a 54 percent lower defect rate3. The most reactive quarter saw 13.0 percent downtime against 4.0 percent for the least reactive, and defect rates of 3.3 against 0.2 percent, though NIST notes many of these differences are not statistically significant3, 8.
DOE estimates that preventive maintenance saves 12 to 18 percent over reactive4. It also cites independent surveys in which functional predictive programs cut maintenance cost 25 to 30 percent and downtime 35 to 45 percent, and eliminated 70 to 75 percent of breakdowns4.
Track the planned share weekly, and keep planned stoppage in a healthy band, generally 3 to 8 percent of scheduled time no cited source. Too little planned downtime signals deferred PMs that will surface as unplanned failures; too much signals over-maintenance that steals available capacity.
How to measure
Planned share: planned and scheduled maintenance hours divided by all maintenance hours, weekly.
Emergency share: hours on emergency jobs divided by total hours worked2.
Planned downtime: planned stoppage divided by scheduled time.
What moves it
Moving work from reactive to preventive and predictive maintenance.
Planning and kitting jobs before they reach the schedule.
Condition data to set the right amount of planned stoppage.
PM compliance is the leading indicator that predicts the others. World-class plants complete 90 percent or more of scheduled PMs within their window2; below 70 percent, MTBF starts sliding within a quarter no cited source.
The cited benchmark is DOE FEMP's preventive maintenance completion percentage, PM actions completed divided by PM actions scheduled, above 90 percent2. The same table sets schedule compliance above 90 percent and maintenance overtime below 5 percent of regular maintenance hours2.
Measure it as PMs completed on time divided by PMs due, and hold the denominator honest by not deferring due dates to flatter the ratio.
The lever is realistic scheduling. If compliance is chronically low, the PM load likely exceeds available labor hours, and you must either add capacity or extend intervals based on condition data rather than skipping work silently.
How to measure
PMs completed on time divided by PMs due.
Never move a due date to flatter the ratio.
Track schedule compliance and maintenance overtime alongside.
What moves it
A PM load that fits available labor hours.
Intervals extended on condition data, never skipped silently.
Added capacity when the load is justified.
Missed PMs show up later as falling MTBF; watch for it with the MTBF calculator.
OEE
Overall equipment effectiveness (OEE) frames availability against quality and speed, and its benchmark is famously demanding. World-class OEE is 85 percent6, 5, built from roughly 90 percent availability, 95 percent performance and 99 percent quality6, 9.
Those factors come with a nuance. Seiichi Nakajima's world-class table, as reproduced by Raju and colleagues, sets them above 90, 95 and 99.9 percent6, while the Lean Enterprise Institute's worked example multiplies 0.90 x 0.95 x 0.99 to get 84.6 percent9.
Typical plants land at 60 percent5, and many score below 50 when they first measure honestly1. Measured data sit close to that line: a Swedish sample of machines averaged 51.5 percent, and published averages range from 40 to 65 percent1.
Do not chase OEE as a single number; decompose it. Availability from your reliability data feeds directly into the first OEE pillar.
For example, a 60 percent score driven by a quality problem needs a completely different fix than the same score driven by availability losses.
For example, closing a 7 point availability gap on a line that runs about 4,000 hours a year and costs 3,000 dollars an hour recovers 0.07 x 4,000 x 3,000, or about 840,000 dollars a year, which outranks a flashy MTBF win on a cheap asset.
Write down each KPI's definition, including what counts as downtime, and keep it fixed.
Collect several months of data per asset before you set targets.
Set stretch targets one tier above your current typical range.
Review the top five bad actors monthly and re-baseline quarterly.
Give every KPI an owner, a target and a next action.
World-Class OEE: Set Targets To Drive Improvement. Vorne Industries (OEE.com). Sections 'What Is World-Class OEE?', 'TPM and World-Class OEE', 'Some Perspective on World-Class OEE'. Accessed September 30, 2026.
Anatomization of the overall equipment effectiveness (OEE) for various machines in a tool and die shop. Journal of Intelligent Manufacturing and Special Equipment, Vol. 3 No. 1, pp. 97-105 (Emerald, CC BY 4.0), DOI 10.1108/JIMSE-01-2022-0004. published 2022. p. 101 (PDF p. 5): sentence "Seiichi Nakajima included the above four “world-class” numbers." and Table 1 "The world-class OEE factor"; p. 103 (PDF p. 7): "worldwide threshold of >85% set by the Japan Institute of Plant Maintenance (JIPM)". Accessed September 30, 2026.
Economics of Manufacturing Machinery Maintenance (NIST AMS 100-34), Section 5.2 Defects. National Institute of Standards and Technology, Engineering Laboratory (D. S. Thomas, B. A. Weiss). published June 2020. Section 5.2 and Table 5.2 'Defects due to Reactive Maintenance', column 'Rate of Defects (percent)', Total rows for NAICS 321-339 (p. 27); Section 7.1 (p. 35). Accessed September 30, 2026.