How Long Does a Mattress Really Last? We Traced the Seven-Year Rule to Its Source

Where the seven-year rule comes from

Ask any mattress store, most sleep sites, or an AI assistant how often to replace a mattress and you will get some version of the same number: seven years, sometimes stretched to eight or ten. We wanted to know who first put a number on it, so we followed the citations backwards. They stop almost immediately, at one place.

The Better Sleep Council's guidance page puts it plainly: "a mattress set that is seven years old has reached the end of the line and needs to be replaced." The page offers signs of wear and general advice. It cites no study, no dataset, and no evidence of any kind for the seven-year figure. Industry

Who is the Better Sleep Council? Its own site answers: "Sponsored by the International Sleep Products Association (ISPA), and supported by experts from industry and academia." ISPA is the mattress manufacturers' trade association. BedTimes, ISPA's own trade magazine, describes the BSC as "the consumer education arm for ISPA." Industry

So the trail ends here: the number every retailer repeats was published by the industry that sells the replacement. That alone does not make it wrong. It does mean the seven-year rule is a marketing position until someone tests it, and it is fair to ask whether anyone has.

The industry keeps score on how often you replace

A detail that rarely makes it into consumer articles: the mattress industry formally tracks the national "replacement cycle," the average number of years between mattress purchases, and reports it in its trade press the way any industry reports a sales metric.

In 2023, ISPA's magazine headlined a BSC survey result: "Mattress Replacement Cycle Drops to 8.3 Years." An ISPA vice president welcomed the trend in the same article, saying "the BSC's and industry's messages about the need for consumers to evaluate the comfort and support of their current mattress are resonating in the marketplace." A shorter cycle means more mattresses sold; the industry says so in its own words, in its own magazine. Industry

The same survey series shows what people actually do, and it does not match the rule. Per the BSC's own research write-up: "Consumers predominantly expect a mattress to last between eight to 10 years, yet nearly half report that they replaced their mattress within seven." The reported average cycle was 9.6 years in 2023. The generational spread is wide: "millennials (ages 25 to 40) wait the shortest time, an average of 6.5 years," while "baby boomers and older generations (ages 56 and up)" are described as the group "who wait the longest at 11.7 years." Industry

Read those two paragraphs together and the shape of the thing is visible. The public number is seven. The industry's own measurement of real behaviour is closer to nine or ten. And the gap between them is tracked, in the industry's own publication, as a metric to be improved.

The studies behind “timely replacement”

There is published, peer-reviewed research on new mattresses, and it is what industry materials point to when the seven-year advice is challenged. Nearly all of it comes from one research group: Bert Jacobson and colleagues at Oklahoma State University, in a series of papers through the 2000s.

The design was similar across the series. Volunteers rated their sleep quality and back discomfort for about a month on their own beds, the beds were then replaced with new medium-firm ones, and the ratings continued. Ratings improved, and Jacobson et al. 2008 concluded that "continuous sleep quality may be dependent on timely replacement of bedding systems." Peer-reviewed

Three things about this series belong next to that conclusion, and none of them requires reading between the lines; all three are stated in the papers themselves.

First, the funding. The support statement in Jacobson et al. 2006 reads: "Support was received from the International Sleep Products Association, the European Bedding Industries Association, and the Better Sleep Council Canada (all non-profit trade associations)." The research the industry cites was paid for by the industry, including the same organization whose seven-year guidance it now supports. Peer-reviewed

Second, the design. These were before-and-after comparisons on small convenience samples, with no control group, no randomization, and no blinding. Everyone knew they had been given a brand-new bed. A person who receives a free new mattress and is asked daily whether they slept better is not a neutral instrument, and without a control group there is no way to separate a real support effect from a new-bed placebo effect. The improvements reported are real data; what caused them is not separable in this design.

Third, the age of the beds. Jacobson et al. 2008 states it in four words: "Average bed age was 9.5yrs." The beds whose replacement produced the improvement were, on average, two and a half years older than the seven-year rule they are cited to support. Even taken entirely at face value, this series is evidence about beds pushing a decade, not beds at year seven. Peer-reviewed

What independent research has actually tested

Independent sleep-surface research exists. It just tested a different question.

The strongest single trial remains Kovacs et al. 2003 in The Lancet: a randomized, double-blind, multicentre trial in patients with chronic low-back pain, comparing firm against medium-firm mattresses. Medium-firm won, on pain in bed, pain on rising, and disability. Note what varied: firmness. Every bed in the trial was new. Peer-reviewed

A 2015 systematic review by Radwan and colleagues in Sleep Health gathered the controlled trials since 2000 and reached a similar place: a "medium-firm mattress and is custom inflated (self-adjusted)" surface came out optimal for sleep comfort, quality and spinal alignment. The reviewed trials classify mattresses by firmness and design. Replacement age is not among the variables, because no trial has made it one. Peer-reviewed

That absence is the finding of this article, so it should be stated carefully. We searched the controlled-trial literature for any study that randomized when a mattress is replaced, or compared outcomes across mattress age under controlled conditions, and found none; the systematic reviews of this literature likewise contain none. That count is our own search of published trials, not a figure from a paper, and we would genuinely welcome a counterexample. Inferred

So the evidence base under "replace at seven years" is: an industry guidance page with no citations, an industry-tracked sales metric, and industry-funded uncontrolled studies of ten-year-old beds. The independent evidence says medium-firm support helps bad backs. Nobody has tested the calendar.

The chemistry corner nobody mentions

This site's usual beat is what mattresses emit, and the replacement question has a chemistry angle that cuts in an unexpected direction: the emission peak belongs to the new mattress.

Volatile organic compound emissions are at their highest when a mattress is new, in the first days and weeks out of the wrapper, and fall steeply from there; our piece on mattress off-gassing walks through the chamber data. A frequent replacement habit means spending more of your sleeping life in that high-emission phase. Aging does its own slow chemistry, and the honest version is not "old beds emit less" but "old beds emit differently": breakdown products form as the foam oxidizes over a service life, an arc we trace in how a mattress ages chemically. But if the seven-year rule were about chemistry, it would point the other way from where the marketing points: the newest bed in the room is the one at its emission peak.

The honest chemical summary: replacement timing is a support and hygiene question, not an emissions question. Nobody should keep a sagging mattress because of VOCs, and nobody should replace a supportive one to escape them.

What this does not prove

Everything above traces the seven-year rule to marketing. It does not prove the rule wrong, and it is worth saying so before a critic does.

Mattresses genuinely wear out. Foam softens and takes a set, springs lose resilience, and a surface that has drifted away from supporting your spine can plausibly do to your back what the firmness trial measured, since a soft old bed and a too-soft new bed present similar surfaces. Allergen and dust-mite load also accumulates with years, which matters for allergic sleepers and is managed with washing and encasements either way. It is entirely possible that, for some construction at some price point, seven years is a fair average. The point is narrower: nobody has measured it, and the only parties publishing a number profit when the number is small.

The test that would settle it is not exotic: randomize households with mattresses of known age to replacement now versus replacement later, blind the outcome collection, and follow sleep quality and pain for a year. Until someone without a mattress to sell funds that study, "replace at seven years" remains a slogan wearing a lab coat.

So when should you actually replace it?

Judge the mattress, not the calendar. The signals that something real has changed:

• You wake with aches that fade once you're up and moving. Pain that is worst at rising and eases within the hour points at the sleep surface rather than your spine.
• Visible sag or a body-shaped hollow that a rotation doesn't fix. Put a straightedge or a broom handle across the surface if you want to be sure your eyes aren't negotiating.
• You consistently sleep better on other beds — hotels, guest rooms — and notice the difference on the first night, not just the novelty. We looked at what's really behind that effect in why you sleep better in hotels; when it's the bed, it's usually support.
• Allergies are worse in bed than anywhere else despite washed bedding and an encasement. Age is one plausible cause among several.

If none of those is true, the age of the mattress is not, on current evidence, a reason to replace it. If one of them is true at year four, the seven-year rule should not talk you out of acting early either. And if you do replace, the choice that has actual trial evidence behind it is support: medium-firm, adjusted to you. For what to check on the way in, from fire barriers to certifications, start with the fiberglass check and what CertiPUR-US actually covers; for the case where the "new" bed is second-hand, our piece on hand-me-down crib mattresses covers when used is genuinely fine.

---

What this article does not establish: an evidence-based replacement interval. We looked for one and it does not exist, in either direction. This piece also does not establish that industry funding falsified the Jacobson results; the studies report what they report, and the design limits are stated in our reading above, not alleged. What we can say with confidence is where the seven-year number comes from, what stands beneath it, and that the honest replacement decision runs on symptoms a sleeper can check in one morning.

This article is part of Embr's chemistry research series. Our methodology and editorial commitments are published openly. Embr sells no mattresses, runs no affiliate links, and accepts no sponsorship — on this topic, that is not a boast, it is the method.