Search “reactive maintenance cost” and you’ll hit the same sentence over and over. Reactive maintenance costs three to five times more than preventive maintenance, according to the U.S. Department of Energy.
It shows up in CMMS marketing, consulting decks, conference slides, and trade articles. The attribution is always the same and always vague: the DOE. No document, no page, no year.
So we went and read the document.
The claim points at one place, the Federal Energy Management Program’s Operations and Maintenance Best Practices: A Guide to Achieving Operational Efficiency. We pulled all three surviving editions and read the chapter where DOE compares the cost of maintenance programs. The phrase isn’t there. What is there is a cost table with a stranger history than the slogan suggests, and the ratios it produces don’t match the number everyone quotes.
This is a compiled reference on where the multiplier comes from, what DOE published, and what you can defend in a budget meeting.
What the Guide Says and Where It Came From
The FEMP guide is real, and it’s the right document to be looking at. It exists in three editions, all credited to the same four authors at Pacific Northwest National Laboratory (G.P. Sullivan, R. Pugh, A.P. Melendez, and W.D. Hunt), all prepared for DOE’s Federal Energy Management Program:
- The original, December 2002, catalogued as PNNL-13890. It carries no release number on its cover.
- Release 2.0, July 2004, catalogued as PNNL-14788.
- Release 3.0, August 2010.
Release 3.0 is still current. PNNL’s O&M Best Practices project page pointed to it as the reference document as of this writing, sixteen years after publication. We found no Release 4.0.
Chapter 5, “Types of Maintenance Programs,” is where the money numbers live. It compares four program philosophies and assigns each an annual cost:
- Reactive maintenance (run to failure): $18 per horsepower per year
- Preventive maintenance (time-based): $13 per horsepower per year
- Predictive maintenance (condition-based): $9 per horsepower per year
- Reliability centered maintenance: $6 per horsepower per year
Those four figures are the numeric spine of the entire folklore. And Chapter 5 gives exactly one reference for them.
It’s Piotrowski, J. April 2, 2001, “Pro-Active Maintenance for Pumps,” Archives, February 2001, Pump-Zone.com. Reprinted with permission of Pump-Zone.com.
That’s a 2001 article in a pump industry trade publication. DOE’s citation gives only Pump-Zone’s homepage, not an article-level URL. Each of the four figures is attributed generically, in the guide’s own words, to studies that have “shown” the cost, with no study named.
The phrase “three to five times” appears nowhere in that chapter, in any of the three editions. Neither does any statement of a general cost multiple between reactive and preventive maintenance.
The Reliable Confidence Score
| Claim | Figure | Reliable Confidence | What It Really Means |
|---|---|---|---|
| DOE says reactive maintenance costs 3 to 5 times more than preventive | 3x to 5x | LowPhrase not found in the cited source | The wording appears in none of the three FEMP guide editions and in no DOE publication we reviewed. The attribution is doing work the document doesn’t support. |
| FEMP’s published maintenance program cost comparison | $18 / $13 / $9 / $6 per hp per year | MediumReal DOE text, trade-press origin | This one is in the guide and has been since 2002, but it’s sourced to a single 2001 pump trade article and expressed as an annual rate per installed horsepower. |
| Reactive-to-preventive ratio implied by DOE’s own figures | About 1.4 to 1 | HighArithmetic from the published table | $18 against $13. If you cite the guide for a reactive-versus-preventive multiple, this is the multiple you are citing. |
| Reactive-to-RCM ratio implied by DOE’s own figures | About 3.0 to 1 | HighArithmetic, but a different comparison | $18 against $6. Among the four published figures, the 3.0 ratio compares reactive maintenance with reliability centered maintenance, not with preventive. The guide states no ratios of its own. |
| Preventive maintenance saves 12% to 18% over reactive | 12% to 18% | MediumGuide text, no study named | In all three editions and widely quoted correctly. DOE attributes it to unnamed studies, so it inherits their uncertainty. |
| Reactive maintenance costs up to 10 times more | 10x | LowNo source located | Circulates alongside the 3-5x claim, also credited to DOE. Nothing in any edition approaches this figure. |
| Reactive-heavy plants and unplanned downtime (NIST) | 10.4% vs 4.9% of planned production time | MediumPeer-reviewed, small sample, author-flagged as anecdotal | A real measured comparison of US manufacturers. It measures downtime and defects, not cost per repair, and NIST cautions against leaning on it hard. |
| Any such statement in the three FEMP guide editions | None found | HighNarrow negative finding, three editions read | Across the 2002, 2004, and 2010 editions, DOE never states a general reactive-to-preventive cost multiple. |
The Big Takeaway
The multiplier didn’t come from nowhere. It came from somewhere and then drifted.
Run the arithmetic on DOE’s own table. Reactive at $18 against preventive at $13 is a ratio of about 1.4 to 1. Reactive at $18 against reliability centered maintenance at $6 is 3.0 to 1. One plausible explanation for the circulating 3x is that the reactive-to-RCM ratio was later repeated as a reactive-to-preventive ratio. We found no documented transmission trail, so that stays an explanation rather than a finding.
The guide’s own numbers put reactive at $18 per horsepower per year and preventive at $13. That’s a ratio of about 1.4 to 1, not 3 to 1.
The “5x” has no ancestor we could find in the document at all. Neither does the “10x” variant.
There’s a second drift, and it matters more than the arithmetic. Dollars per horsepower per year is an annualized rate across a population of motor-driven equipment. It is not the cost of a repair. When a vendor writes that reactive maintenance “costs 3 to 5 times more per repair” and credits DOE, the unit has changed underneath the number. The source measures a program’s annual cost per unit of installed capacity. The claim measures a job.
Why the Numbers Vary
Part of the variation is that the underlying figures never moved while everything around them did.
We read Chapter 5 in all three editions. The dollar figures are identical across December 2002, July 2004, and August 2010: $18, $13, $9, and $6. The citation is identical: Piotrowski 2001. Release 3.0’s preface states that it “provides updates to Release 2.0 in the areas of O&M technologies, equipment performance, and costs.” The cost table was not among the things updated. A 2001 trade-press number passed through a stated cost revision in 2010 unchanged, and it has now been the federal government’s published maintenance cost comparison for twenty-three years.
One thing did change. In the 2002 and 2004 editions, the comparison carries the header “Case Study Comparison of Four Maintenance Programs (Piotrowski 2001).” In Release 3.0, the same block reads “Comparison of Four Maintenance Programs (Piotrowski 2001).” The words “Case Study” are gone. A figure labeled as one case’s experience through two editions reads as a general comparison in the third.
The rest of the variation is downstream. Once a number loses its unit and its scope, every writer who repeats it rounds in whichever direction suits the sentence. We found “3 to 5 times more expensive than preventive,” “3-5x more per repair,” “three to five times more than planned preventive maintenance,” and “up to 10 times more,” all in currently published pages, all crediting DOE. One vendor page states that the 3-5x multiplier “is not an estimate” but “a consistently measured ratio across manufacturing maintenance cost studies,” without naming a study.
Not everyone repeating it presents it as a DOE finding. A Schneider Electric post from 2014 calls the reactive-versus-preventive premium “a generally-accepted industry rule of thumb” of three to four times, explicitly labels it a rule of thumb, and points readers to Campbell and Jardine’s Maintenance Excellence: Optimizing Equipment Life Cycle Decisions (Marcel Dekker, 2001), which proposes assigning separate cost units to planned work, unplanned work, and emergency breakdown work. That framing is honest about what the number is. Klaus Blache of the University of Tennessee’s Reliability and Maintainability Center gives a wider range still, writing that emergency repairs can cost three to more than ten times planned maintenance. Practitioner ranges like these are useful, and they are not the same thing as a federal benchmark.
What to Cite Instead
If you need a measured comparison between reactive-heavy and proactive-heavy operations, there is one, and it isn’t the FEMP table.
Douglas Thomas and Brian Weiss at the National Institute of Standards and Technology surveyed US manufacturers and published the results as NIST Advanced Manufacturing Series 100-34 in 2020. A related 2021 paper in the International Journal of Prognostics and Health Management expands on that work using Monte Carlo analysis and resampling.
The two produce different headline totals, and they’re worth keeping apart. The 2020 report put total maintenance-related costs and losses in the industries studied at $193.6 billion for 2016, of which $119.1 billion was losses rather than maintenance spending proper. The 2021 paper’s Monte Carlo analysis puts the total at an average of $222.0 billion.
The comparison is the useful part, and it’s worth quoting with its structure intact. NIST split respondents by maintenance expenditure, putting the top half for reactive reliance in one group and the rest in another. The reactive-heavy group reported 68.8% of its machinery maintenance and repair costs as reactive, against 22.2% for the other group. Unplanned downtime consumed 10.4% of planned production time for the reactive-heavy group against 4.9% for the others, a reduction of 52.7%. Defect rates were 78.5% lower for the less reactive group.
The 2020 report ran a different cut that produces the figure most often seen in circulation. Establishments in the top quarter for reactive maintenance reliance were associated with 3.28 times more downtime than those in the bottom quarter, 13.0% of planned production time against 4.0%, along with 16 times the defect rate. Quartile against quartile is a sharper contrast than half against half, which is why those multiples are larger. If you use that figure, carry the comparison group with it. Most secondary coverage doesn’t.
The report is careful about what that comparison will bear. It states that given the small groupings, many of the differences are not statistically significant and should be read as anecdotal evidence. Two survived a t-test at the 10% level: downtime specifically attributed to maintenance issues, and the increase in inventory. The headline 3.28 figure is not among them.
Read the units carefully, because this is where the FEMP claim goes wrong and NIST does not. These are downtime and defect comparisons, not cost-per-repair comparisons. NIST measured how much more downtime reactive-heavy plants experience. It did not measure how much more a reactive repair costs than a planned one.
Two caveats belong with the figures, and NIST states both. The survey drew 85 responses and used 71 after removing incomplete and out-of-scope submissions. And the authors describe the grouped comparisons, including the reactive one, as evidence that “should be treated as anecdotal” because the groupings leave small numbers in each cell. That is unusually candid for a government study, and it’s why this article rates the finding Medium rather than High. It’s the best measured comparison we found. It is not a settled benchmark.
There’s a finding in the same work that cuts against the easy version of the story. The less reactive group spent 81.7% more on direct maintenance relative to shipments. Spending less on maintenance and losing less to downtime are different outcomes, and a proactive program can raise the first while lowering the second.
How to Use the Numbers Safely
If you’re building a business case, three moves keep you out of trouble.
Cite the figure you’re using, with its unit. “DOE’s FEMP O&M guide puts reactive maintenance at $18 per installed horsepower per year against $13 for preventive” is defensible and checkable. “DOE says reactive costs 3 to 5 times more” is neither.
If you need a multiple, calculate your own. You have the work order history. Compare the fully loaded cost of emergency work at your site, including overtime premiums, expedited freight, and lost production, against the same scope executed on schedule. A number you measured beats a number you inherited, and it survives the question “where did that come from.” Our guide to planned versus reactive maintenance spend ratios covers how to set the categories and the cost boundary before you start counting, which is where most internal calculations come apart.
If you’re quoting the 12% to 18% savings figure, quote it as what it is. It’s in all three editions, it’s the most frequently correct citation in this whole cluster, and DOE attributes it to unnamed studies. Present it as DOE’s published estimate rather than a measured result.
Where Teams Go Wrong
The most common failure is unit substitution. Someone reads a per-horsepower annual program rate and repeats it as a per-repair cost. The number survives the trip; the meaning doesn’t.
The second is treating federal publication as validation of the underlying research. DOE printed the Piotrowski figures and attributed them. The guide does not document that DOE generated or independently validated them, or that they are representative. Federal publication does not, by itself, turn the underlying trade-press figures into a federal study.
The third is the credibility trade. Quoting “3 to 5 times, per DOE” in a room with a reliability engineer in it is a risk, because the guide is short, free, and one search away. Getting caught inflating a source costs more than the slide gained.
Methodology
We fetched and read Chapter 5 of all three FEMP editions in full: PNNL-13890 (December 2002), PNNL-14788 (Release 2.0, July 2004), and Release 3.0 (August 2010), along with the front matter and Chapters 1 through 4 of each, and compared them directly. Every dollar figure and reference quoted above is from those documents. On the NIST side, the 50/50 figures come from Tables 5 and 6 of the 2021 paper; the quartile figures come from the abstract and Table 7.2 of AMS 100-34; and the wording of how respondents apportioned maintenance is from question 7 of the Machinery Maintenance Survey, reproduced in Appendix A of AMS 100-34. None of the NIST figures here are taken from press coverage.
We searched for the “three to five times” phrasing across DOE and FEMP materials and did not find it. That is a negative finding across the sources reviewed rather than a claim that no such DOE sentence exists anywhere. Chapter 5 is where maintenance program costs are compared in all three editions, and it isn’t there. We did not review the work-management standards for this article; our companion guide on planned versus reactive maintenance spend ratios covers SMRP’s metrics and EN 15341:2019+A1:2022 and reports finding no universal planned-versus-reactive spending benchmark in either.
Confidence tiers reflect confidence in the claim, not enthusiasm for the number. A narrow negative finding can rate High. A real figure with a thin pedigree rates Medium. A figure we could not source at all rates Low, even where it’s repeated constantly. A peer-reviewed figure whose own authors flag it as anecdotal rates Medium, not High.
Frequently Asked Questions
Does the U.S. Department of Energy say reactive maintenance costs 3 to 5 times more than preventive maintenance?
Not in the three FEMP guide editions normally cited for the claim. The phrase does not appear in the December 2002, July 2004, or August 2010 editions of DOE’s Federal Energy Management Program O&M Best Practices Guide. Chapter 5 compares four maintenance programs by annual cost per horsepower and never states a general reactive-to-preventive multiple.
What does the DOE FEMP guide publish for maintenance program costs?
Reactive maintenance at $18 per horsepower per year, preventive at $13, predictive at $9, and reliability centered maintenance at $6. All four figures are cited to a single 2001 article by J. Piotrowski in the pump trade publication Pump-Zone.com, and all four are identical in the December 2002, July 2004, and August 2010 editions.
What ratio do DOE’s own numbers produce?
Reactive against preventive is about 1.4 to 1 ($18 versus $13). Reactive against reliability centered maintenance is about 3.0 to 1 ($18 versus $6). The 3x figure in circulation matches the reactive-versus-RCM comparison rather than reactive versus preventive.
How many editions of the FEMP O&M Best Practices Guide are there?
Three. The December 2002 original (PNNL-13890), Release 2.0 (July 2004, PNNL-14788), and Release 3.0 (August 2010), which remains current. Release 3.0 states it updated Release 2.0 on costs, but the four per-horsepower figures are identical in all three.
Is there a measured comparison between reactive and proactive maintenance?
Yes. Thomas and Weiss at NIST surveyed US manufacturers and found that establishments in the top half for reactive maintenance reliance lost 10.4% of planned production time to unplanned downtime, against 4.9% for the rest, with defect rates 78.5% lower in the less reactive group. The authors flag the grouped comparisons as anecdotal given the sample of 71 usable responses.
Why does the NIST study rate Medium and not High?
Because its authors say so. The paper describes the grouped comparisons as evidence that should be treated as anecdotal, since splitting 71 respondents into groups leaves small numbers in each. The finding is directionally informative and the best available, and it is not a settled benchmark.
Where does the 12% to 18% preventive maintenance savings figure come from?
It is published in all three editions of the FEMP O&M Best Practices Guide as the estimated cost saving of a preventive maintenance program over a purely reactive one. DOE attributes it to unnamed studies, so it is best presented as DOE’s published estimate rather than a measured finding.
The Short Version
The FEMP guide is a real DOE document and worth reading. It just doesn’t say what it’s constantly credited with saying.
Its numbers put reactive maintenance at $18 per horsepower per year and preventive at $13, a ratio of about 1.4 to 1. The 3x in circulation matches the reactive-versus-RCM comparison instead. The 5x and the 10x match nothing in the document. And all four figures trace to one 2001 pump trade article that has been reprinted without revision across three editions and twenty-three years.
Reactive maintenance does cost more. NIST measured reactive-heavy plants losing more than twice the share of planned production time to unplanned downtime, and said plainly that the grouping is small enough to treat as anecdotal. Cite that with its caveat, cite the FEMP dollar figures with their unit attached, or measure your own multiple from your own work orders.
Just stop citing DOE for three to five times. The guide everyone means doesn’t say it.
Sources
- U.S. Department of Energy, Federal Energy Management Program, Operations & Maintenance Best Practices: A Guide to Achieving Operational Efficiency, Release 3.0, August 2010 (G.P. Sullivan, R. Pugh, A.P. Melendez, W.D. Hunt; Pacific Northwest National Laboratory): https://www.energy.gov/sites/prod/files/2020/04/f74/omguide_complete_w-eo-disclaimer.pdf
- FEMP O&M Best Practices Guide Release 3.0, Chapter 5, “Types of Maintenance Programs”: https://www1.eere.energy.gov/femp/pdfs/om_5.pdf
- U.S. Department of Energy, Federal Energy Management Program, O&M Best Practices: A Guide to Achieving Operational Efficiency, Release 2.0, July 2004, PNNL-14788: https://www.pnnl.gov/main/publications/external/technical_reports/pnnl-14788.pdf
- U.S. Department of Energy, Federal Energy Management Program, Operations & Maintenance Best Practices: A Guide to Achieving Operational Efficiency, December 2002, PNNL-13890: https://www.pnnl.gov/main/publications/external/technical_reports/PNNL-13890.pdf
- OSTI record, December 2002 edition (PNNL-13890), DOI 10.2172/15010224: https://www.osti.gov/biblio/15010224
- OSTI record, O&M Best Practices Guide Release 2.0: https://www.osti.gov/biblio/947512
- OSTI record, O&M Best Practices Guide Release 3.0: https://www.osti.gov/biblio/1220381
- Pacific Northwest National Laboratory, O&M Best Practices project page: https://www.pnnl.gov/projects/om-best-practices
- U.S. Department of Energy, Federal Energy Management Program, “Operations and Maintenance Challenges and Solutions”: https://www.energy.gov/cmei/femp/operations-and-maintenance-challenges-and-solutions
- Thomas, D. and Weiss, B. (2020), Economics of Manufacturing Machinery Maintenance: A Survey and Analysis of U.S. Costs and Benefits, NIST Advanced Manufacturing Series 100-34, DOI 10.6028/NIST.AMS.100-34: https://nvlpubs.nist.gov/nistpubs/ams/NIST.AMS.100-34.pdf
- Thomas, D. and Weiss, B. (2021), “Maintenance Costs and Advanced Maintenance Techniques in Manufacturing Machinery: Survey and Analysis,” International Journal of Prognostics and Health Management, 12(1), DOI 10.36001/ijphm.2021.v12i1.2883: https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=931612
- Blache, K.M., “Twenty Reasons to Stop Firefighting,” University of Tennessee Reliability and Maintainability Center: https://rmc.utk.edu/twenty-reasons-to-stop-firefighting-dr-klaus-m-blache
- Schneider Electric, “Preventive Maintenance: Pay Now or Pay (More) Later,” September 2014, as an example of the claim presented as an industry rule of thumb: https://blog.se.com/energy-management-energy-efficiency/2014/09/04/preventive-maintenance-pay-now-pay-later/
- Campbell, J.D. and Jardine, A.K.S., Maintenance Excellence: Optimizing Equipment Life Cycle Decisions (Marcel Dekker, 2001)
- Examples of the claim in current circulation, attributed to DOE: TrellisPoint, eWorkOrders, Oxmaint, Fabrico, Opsima
