Our method
How we test a mattress
Published in full, including the parts that make us look slow and expensive, so you can check our work and decide for yourself whether a score means anything.
- Mattresses tested
- 43 since 2021
- Currently published
- 9 reviews
- Nights slept
- 273 in this round
- Minimum per bed
- 30 nights
Pressure mat, surface probes and the dial gauge
How we choose what to test
We start from search demand and reader email, not from affiliate programmes. Each quarter we list the mattresses people are actually asking about, cross out the ones we have already measured in the last eighteen months, and buy the rest. Where a manufacturer offers a review unit, we accept it only on the condition that there is no editorial oversight, no advance sight of the review, and no obligation to publish at all.
Roughly one in five mattresses we test never gets a review. If a bed is unremarkable and beaten on every axis by something we have already recommended, publishing a review of it is a service to the affiliate programme rather than to you.
The room and the equipment
Testing happens in one bedroom, held at 68°F and 50% relative humidity by a thermostat and a dehumidifier, on a solid platform base with no slat gaps. The bedding is identical for every mattress: percale cotton sheets, a 7-tog duvet, and the same two pillows. Changing any of those would change the temperature results by more than the difference between mattresses.
The instruments are unglamorous and they are the same ones every time:
- A digital compression gauge for firmness, sampled at nine points across the surface
- A 1,024-sensor pressure mat for shoulder, hip and heel loading
- Three thermocouple surface probes with an eight-hour logger
- A triaxial accelerometer and a 10 lb calibrated drop weight
- A dial indicator on a straight edge for sag and edge compression
The seven-part rubric
Every mattress is scored out of ten on the same seven measures, weighted as below. The overall score is the weighted mean, rounded to one decimal place. It is not adjusted, and it is never changed for commercial reasons — though it does change when a price changes, because value is part of it.
- SupportWeight 25%
- Spinal alignment photographed from the side on the back and on the side, with markers at the shoulder, hip and knee. Deviation from neutral is measured in degrees. Hip sink is measured with a dial gauge under each tester.
- Pressure reliefWeight 20%
- A 1,024-sensor pressure mat records peak and average pressure at the shoulder, hip and heel in three positions. Scores are normalised against the test-set average so they stay comparable across rounds.
- TemperatureWeight 15%
- Three surface probes at shoulder, hip and knee, logged hourly for eight hours under a sleeping tester in a room held at 68°F with 50% relative humidity and identical bedding.
- Motion transferWeight 12%
- A 10 lb weight dropped from 12 inches at the centre of the mattress, with an accelerometer at the far edge. Repeated ten times; the median peak acceleration is recorded.
- Edge supportWeight 10%
- A tester sits on the perimeter for ten minutes with a dial gauge recording compression at one minute and at ten, so we catch mattresses that keep giving way rather than settling.
- DurabilityWeight 10%
- Foam density measured on cut samples, coil gauge verified, and centre-of-bed sag re-measured every twelve months on every mattress we have kept since 2021.
- ValueWeight 8%
- Total score against the price actually paid for a Queen, adjusted for trial length and warranty terms. This is the only sub-score that changes when a price changes.
The testers
Firmness is relative to the person lying on it, so a single tester produces a single-person review. Ours are 135 lb (side sleeper), 175 lb (back sleeper) and 240 lb (combination). Each sleeps on every mattress for at least seven consecutive nights, and the primary tester carries the remaining nights to thirty.
Where the three disagree — and on soft mattresses they always do — we publish the disagreement rather than averaging it away. That is why almost every review on this site includes a weight range in the “look elsewhere if” list.
Where the money comes from
Affiliate commission, and nothing else. There is no display advertising, no sponsored content, and no paid placement anywhere on this site. When you buy through one of our links the retailer pays us a percentage, and that percentage is the same whether we rank a mattress first or last — which is precisely why we can afford to rank it last.
The commission rates we are offered are not equal across brands. We publish the full disclosure, including which programmes we are members of, so you can judge for yourself whether our rankings track the money. They do not: our top-rated mattress pays us less than the third-placed one.
When we get it wrong
We have changed a published score four times since 2021 — twice after a manufacturer altered the construction without changing the model name, once after our own measurement error, and once after enough reader reports about a failure mode we had not seen in thirty nights.
Every change is logged at the foot of the review with the date and the reason. If you think something here is wrong, tell us and include the detail. Corrections are published, not quietly edited.
Questions about our testing
Do brands pay to be included or ranked higher?
What happens to the mattresses after testing?
Who are the testers?
How often are reviews updated?
See the method applied
Browse every review, start from a problem you are solving, or read our editorial policy.