Blog / Reliability
Reliability · Guide

What Is a Good MTBF? How to Set Your Own Target

OL By OEE Lab Editorial |Updated 2 November 2026 |6 min read
Short answer

A good MTBF is one that keeps failure losses inside what your line can afford, and that beats your own baseline. We found no neutral source for a universal or world class MTBF, because the number depends on the machine, its duty and what you count as a failure. Set the target with this formula: required MTBF = MTTR x A / (1 - A), where A is the availability you need from failures.

Why a single good number cannot exist

MTBF is operating time divided by failures, so it depends on what you call a failure. The IEC definition fixes the time as operating time, but your plant still has to define the event.

Example: one filler runs 1,000 operating hours. Here is the same machine under three counting rules.

What you countEventsResult
Stops that need a technician10100 h
Any stop over 5 minutes that needs a person5020 h
Every stop over 1 minute4002.5 h
Hypothetical filler. Same machine, same hours, and a 40-fold spread. The last row is closer to mean time between stops than to MTBF.

So any table of "good MTBF" means little unless it says how failures were counted. It also depends on duty: speed, product, shifts and machine age all move the figure.

We found no neutral published table of good MTBF by machine type. If you find one, check its source, its failure definition and who is selling what next to it.

What you can compare against instead

  • Your own history. The same machine, under the same definition, is the fairest comparison.
  • Your sister machines. Two similar fillers on similar products show what is possible in your plant.
  • Published loss figures. These are about time lost, not MTBF, but they are neutral.

In a NIST survey of US discrete manufacturers (2016 data), unplanned downtime averaged 7.8% of planned production time. The quarter of plants that relied most on reactive maintenance lost 13.0%, and the quarter that relied least lost 4.0%.

That covers all unplanned downtime, not only failures, in one US sector. Use it as a rough sense of scale, and never as an MTBF goal.

Method: set the target in five steps

1
Write the failure definition
Example: an unplanned stop that needs a technician, a part or a repair.
2
Get a baseline per machine
Use operating hours and failures for 6-12 months, not one good month.
3
Set a loss budget
Decide how much planned time failures may take on this machine.
4
Work out the required MTBF
Use your real MTTR in the formula below.
5
Check it against your failure list
Name the failures you plan to remove. If you cannot, the target is a wish.
Top-down from the loss budget, then bottom-up from the failure list. The target is where they meet.

From a loss budget to an MTBF target

Availability from failures alone is MTBF / (MTBF + MTTR). Solve that for MTBF and you get the target directly.

Required MTBF = MTTR x A / (1 - A)
MTTR 3 h, A 97%: 3 x 0.97 / 0.03 = 97 h
MTTR 2 h, A 97%: 2 x 0.97 / 0.03 = 64.7 h
A is the availability you need from failures only. Other stops, such as changeovers, take their own share.

Notice the second line of the example. Cutting repairs from 3 hours to 2 lowers the MTBF you need by a third, so the two targets should be set together.

Availability neededMTTR 1 hMTTR 2 hMTTR 4 h
95% (failures take 5%)19 h38 h76 h
97% (failures take 3%)32.3 h64.7 h129.3 h
99% (failures take 1%)99 h198 h396 h
This is arithmetic, not a benchmark. It shows what your own budget requires.

Worked example: one machine, one target

Example: a filler has 20 failures in 1,200 operating hours, and each takes 3 hours to restore. Its MTBF is 1,200 / 20 = 60 hours.

Failures take 60 hours, so planned time is 1,260 hours. Availability from failures is 1,200 / 1,260 = 95.2%.

The failure log shows 12 of the 20 come from two repeat problems. The team expects to remove 8 of those, which leaves 12 failures and an MTBF of 1,200 / 12 = 100 hours.

Failures now take 36 hours. For the same 1,200 operating hours, availability from failures rises to 1,200 / 1,236 = 97.1%, a gain of 1.9 points.

At 90% performance and 98% quality, those 1.9 points are worth about 1.6 points of OEE. So the target reads: "MTBF 100 hours on the filler within 12 months, by removing two repeat failures".

That target has a reason, an owner and a list behind it. Check it against the wider plan in how to set a downtime target and how to set OEE targets.

Do not trust an MTBF from a few failures

A small count swings hard. Example: the same point estimate of 100 hours, from two amounts of data.

5 failures in 500 h48-254 h
20 failures in 2,000 h69-151 h
Each bar shows the 90% confidence range for the true MTBF on a scale of 0-260 hours. It assumes a constant failure rate.

With 5 failures, the true MTBF could plausibly sit anywhere from 48 to 254 hours. With only 2 or 3 failures, one extra failure can look like a collapse.

The ranges use the standard chi-square method for a constant failure rate, for a log that ends on a date and not at a failure. If it ends exactly at the last failure, the low ends rise to 55 and 72 hours.

In practice, use longer windows for machines that rarely fail, and do not judge a fix on one month.

Common traps with MTBF targets

  • One MTBF for the whole plant. It hides the bad machine. Set targets per machine, and for groups read MTBF for multiple machines.
  • Using calendar hours. MTBF counts operating time, and calendar hours make a one-shift machine look far better than it is.
  • Using the maker's figure as your goal. A datasheet MTBF comes from its own stated conditions, which may differ from your duty.
  • Chasing MTBF by relabelling. Moving a failure to another code raises the number and saves no output.
  • Ignoring OEE. Running slower may raise MTBF while performance falls. Keep OEE beside it.

Start with your own figures in the MTBF and MTTR calculator. The repair side has its own page: what is a good MTTR.

See it live in Fabrico

A target needs a baseline you trust. Fabrico takes stops live from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, operators tag the cause at the line, and you see OEE per line and shift. See how Fabrico tracks stops per machine.

What people ask next

What is a world class MTBF?

We found no neutral source for a world class MTBF figure. The number changes with the machine, its duty and the failure definition. Compare with your own history and similar machines, then set the target from your loss budget.

Is higher MTBF always better?

Yes for one machine under one definition, as long as output holds. If MTBF rises because the line runs slower, OEE may fall. Keep availability, performance and quality beside it.

How many failures do I need for a reliable MTBF?

More is better, and a handful is only a rough signal. With 5 failures the 90% range on a 100-hour MTBF runs from about 48 to 254 hours. Use 6-12 months of data on machines that rarely fail.

What is a good MTBF to MTTR ratio?

The ratio sets availability from failures: a ratio of 32 to 1 is about 97%, and 99 to 1 is 99%. Pick the availability you need, then the ratio follows. Try your numbers in the MTBF and MTTR calculator.

Should stops under a few minutes count in MTBF?

Usually not. They belong in performance or in mean time between stops. See minor stops vs breakdowns for a test you can use.

See OEE per line and shift on your own machines

Fabrico is a light MES for OEE. It takes stops, speed and counts from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and operators tag the cause at the line.

Book a Fabrico demo

← More OEE Lab guides · How to improve MTBF →

OEE essentials: OEE calculator · What is OEE · Six big losses · Availability · TEEP · Cost of downtime · OEE benchmark · Free OEE tracker · Free OEE software · How to track downtime · How to implement OEE · Light MES vs full MES