Firmness, support and sleeping position: how to choose a mattress

Support is the core, firmness is the first inch

The comfort layer is what your hand touches in a store: quilting, foam, a pillow top. It compresses under local pressure and it is what people mean by soft or firm. The support core underneath — pocketed springs, high-density foam, latex — is what resists the heaviest part of your body sinking out of line. These two layers are specified separately by manufacturers and sold together as a single adjective.

A useful test is to lie on your side and have someone look along the line of your spine. If it runs straight from neck to tailbone, the core is holding your hip and shoulder where they belong. If the hip has dropped and the line bends, the core is too soft for your weight regardless of how firm the top felt when you pressed it with your hand.

This is also why two people of different weights report opposite things about the same bed. The same core compresses further under more weight, so a medium that supports a 140 lb sleeper can behave like a soft one under a 220 lb sleeper. Firmness labels describe the surface, not the sleeper, and no manufacturer's scale is calibrated to your body.

What the fill actually is, and how to read the label

A mattress sold as memory foam, hybrid or latex is describing the comfort layer in most cases, and the core may be something else entirely. Foam is specified by density and by indentation force deflection; springs by coil count, gauge and whether they are pocketed; latex by whether it is natural, blended or synthetic. A listing that gives an adjective and no number has not told you what you are buying.

Textile labelling helps more than people expect on the fabric side. Under the Federal Trade Commission's guidance on the Textile and Wool Acts, fibres making up 5 percent or more of a product's weight must be named with their percentage in descending order of weight, with anything under 5 percent grouped as other fibres unless it has a definite functional significance, such as 4 percent spandex in a stretch cover. Wool must always be named whatever the proportion.

Use that rule when comparing covers and bedding. A cover described only as a blend has told you nothing; one that reads 65 percent polyester, 35 percent viscose has told you the whole composition. The same rule governs the sheets you will buy to go on top, which the budget guide on this site costs out.

• Foam: look for a density figure and an indentation force deflection figure, not the word plush
• Springs: coil count alone is weak; gauge and whether coils are individually pocketed matter more
• Latex: natural, blended and synthetic behave differently and cost differently
• Covers and bedding: fibres at 5 percent or more must be named with a percentage
• Any fibre under 5 percent appears as other fibres unless it is functional, like 4 percent spandex

The flammability standards tell you something, and not what you think

Two federal standards apply. 16 CFR Part 1632 covers smouldering ignition and requires that a mattress or mattress pad must not char more than 2 inches in any direction when exposed to a lit cigarette. 16 CFR Part 1633 covers open flame and sets two limits for a mattress set: peak heat release must not exceed 200 kilowatts at any point in a 30 minute test, and total heat released must not exceed 15 megajoules in the first 10 minutes.

That is a fire performance floor that every mattress legally sold has to meet, and it is the reason a compliance label is attached to the product. It says nothing at all about comfort, durability, foam density or how long the bed will last, and a listing that leads with compliance as though it were a quality claim has changed the subject.

Where it does help is as a sanity check on an unmarked import. A mattress with no prototype identification and no compliance label is a mattress whose seller has not shown it meets a standard it legally must meet, and that is a reason to stop reading rather than a detail to resolve later.

Test it properly, and know what the trial period is for

A 2 minute sit on the edge of a bed in a showroom reports on the quilting. Lie down in the position you actually sleep in, with a pillow of roughly the height you use at home, and stay there for at least 15 minutes. The core only starts telling you anything once your weight has had time to settle into it, and the difference between two models is usually obvious by minute 10 and invisible at minute 2.

A home trial is a different instrument. Foam and latex both change feel over the first few weeks as the material relaxes, and a body takes time to stop reporting novelty as discomfort. A trial long enough to cross that period is worth more than any in-store comparison, and the conditions that matter are the return cost, who collects it and whether the trial restarts if you exchange rather than return.

Finally, test the bed with the base you will put it on. A pocketed spring unit on slats spaced too far apart behaves like a softer bed and voids most manufacturers' warranties at the same time; the maximum slat gap is published in the care documentation that comes with the mattress, and it is usually the single cheapest thing to get wrong.

Before you buy

• Separate the question into support and firmness before you look at any product
• Lie in your own sleeping position for at least 15 minutes, not 2
• Have someone check that your spine line runs straight while you lie on your side
• Ask for the foam density and indentation figures, or the coil gauge, rather than the adjective
• Read the fibre percentages on the cover and the bedding, remembering the 5 percent naming rule
• Confirm the mattress carries its flammability compliance label and prototype identification
• Measure the slat gap on the base you own against the maximum in the mattress documentation

Questions readers ask

Is a firm mattress better for back pain?

Not by default. What helps is a core that holds your hips and shoulders in line, and on a side sleeper that often means a surface soft enough to let the shoulder in. A very firm surface on a side sleeper tends to push the shoulder up and bend the spine the other way, which is the opposite of the intended effect.

How long should a mattress last?

Long enough that the purchase price stops dominating the cost of a night's sleep, which the budget guide on this site puts at around 8 years for a mid-priced queen. What ends it early is usually the comfort layer compressing into a permanent dip rather than the core failing, and a mattress protector and regular rotation are what slow that down.

Do coil counts tell me anything?

Only alongside gauge and construction. A high coil count made of thin wire is not stiffer than a lower count in heavier gauge, and individually pocketed coils change how motion travels across the bed more than the raw number does. Two beds with the same count can behave completely differently.

Does the compliance label mean the mattress is good quality?

No. It means the product has been shown to meet 16 CFR Parts 1632 and 1633, which are fire performance floors every mattress legally sold must clear. Durability, foam density and comfort are not covered by either standard, and the label is not evidence about any of them.

Can two people with different weights share one mattress comfortably?

Usually yes, with a core that resists the heavier sleeper pulling the lighter one across. Zoned pocketed spring units and split cores are both built for this. The gap worth taking seriously is around 60 lb or more, because that is where the same core compresses noticeably differently under each sleeper.

Sources

1. U.S. Consumer Product Safety Commission — business guidance on mattresses, 16 CFR Parts 1632 and 1633 — checked 2026-11-03
2. Federal Trade Commission — Threading Your Way Through the Labeling Requirements Under the Textile and Wool Acts — checked 2026-11-03