For most people: an N95 or FFP2 with head straps, sealed to your face. It is rated to let in no more than 1 in 10 airborne particles, and with a good seal usually lets in far fewer. A loose surgical mask lets in about 1 in 3, mostly round the edges.
Protection factor is how many times less you breathe in than with no mask: at 10×, 1 in every 10 particles in the air gets in.
Commuting, flights, shops, crowded rooms, a cold going round, wildfire smoke. Folded three-panel and moulded-cup shapes seal on the most faces. About $1 each; reuse one until it gets damp or dirty or the straps go slack.
Caring for someone who is ill, hospital visits, an outbreak, days of smoke. A half mask with P100 or P3 filters costs about $40, usually seals better than a disposable and lasts for years.
Stubble under the edge breaks the seal of any tight-fitting mask. A powered hood (PAPR) blows filtered air in and doesn't need a seal.
It filters your breath on the way out as well as on the way in. An exhalation valve lets some of your breath out unfiltered.
All of these stop particles: dust, smoke, droplets, viruses and bacteria. None of them stop gases such as carbon monoxide.
Four ideas explain almost everything about masks.
It compares the air outside the mask with the air you breathe inside it.
protection factor = particles outside ÷ particles inside
A 10× mask lets in a tenth: 10 particles out of every 100. Going from no mask to 10× keeps out 90 of them. Going from 10× to 20× keeps out only 5 more, but that still halves what you breathe, which matters when the exposure is long or heavy.
Air takes the easiest path in: through the filter, or through gaps where the mask meets your face. Filters are very good. An FFP2 filter must stop at least 94 particles in 100 in the lab, and a P3 filter 9,995 in 10,000.
Yet a half mask with P3 filters is rated only twice as protective as an FFP2. The filter gets over 100 times better; what slips past the seal barely changes. That is why an FFP3 that gaps can protect you less than an FFP2 that seals.
A fit test samples the air in the room and the air inside your mask while you breathe, talk and move your head. The ratio is your fit factor: for that mask, on your face.
fit factor = particles in the room ÷ particles inside the mask
To pass, a disposable respirator or half mask needs a fit factor of at least 100: ten times an N95’s 10× rating. The margin allows for straps slipping and talking in real life. Faces differ: one mask can score 300 on one person and 20 on another, so if a model leaks on you, try a different shape.
1 in every 200 room particles gets inside.
What you breathe in builds up the longer you stay. A mask divides it: 8 hours in a 10× respirator adds up to the same as 48 minutes with no mask. In a surgical mask, it is about 2 hours 40 minutes.
Two things speed the clock up: more particles in the air, as in a crowded, stuffy room or smoke, and breathing harder, as when exercising, singing or talking loudly.
Open the exposure clock →Each bar is the same dose shown as time with no mask, on a 12-hour scale.
Fit matters more than the label. Head straps and a seal check do most of the work.
Share of over 18,000 fit tests that passed, pooled from 29 studies (Knobloch et al., Journal of Hospital Infection, 2023). Every design uses a similar filter.
Top strap high on the back of your head, bottom strap round your neck below your ears.
Press from the bridge of your nose outwards. Pinching with one hand leaves gaps at the sides.
Cup your hands over the mask and breathe out hard. Air at the edges or fogged glasses means it leaks: adjust and try again. Breathing in, the mask should pull in slightly.
Stubble, hair or a scarf under the edge lets air bypass the filter.
Make breathing out easier, but let some of your breath out unfiltered. Skip them if you might be ill.
Braided elastic straps are kinder to long hair than flat ones. If one shape leaks or rubs, try another: faces differ more than masks do.
When the straps go slack, or it is damp, dirty, damaged or harder to breathe through.
Widely available models. We don't sell anything or earn from these links.
N95, FFP2 or KN95. Start with a three-panel or cup shape: they seal on the most faces.
Rubber or silicone, with replaceable filters
N95: “NIOSH” and an approval number (TC-84A-…) printed on the mask. You can look the number up on the CDC’s list.
FFP2 and FFP3: “EN 149” and a CE mark followed by a four-digit number.
KN95: quality varies widely. Buy from known makers and sellers.
Skip FFP1 “dust masks”. Rated 4×, they are too leaky for smoke or infections.
Each country has its own standard. The classes on each row are broadly equivalent for everyday particles.
| Where | Everyday class | Higher class |
|---|---|---|
| EU and UKEN 149 | FFP2≥ 94% | FFP3≥ 99% |
| United StatesNIOSH | N95≥ 95% | N99, N100≥ 99%, 99.97% |
| ChinaGB 2626 | KN95≥ 95% | KN100≥ 99.97% |
| South KoreaKMOEL | KF94≥ 94% | KF99≥ 99% |
| Australia and NZAS/NZS 1716 | P2≥ 94% | P3≥ 99.95% |
| JapanJMHLW | DS2≥ 95% | DS3≥ 99.9% |
Percentages are the minimum share of test particles the filter must stop. A better filter barely helps a disposable: N99 and N100 masks are still rated 10×, because leaks at the seal set the limit.
Protection factors are assigned ratings: the protection an employer can count on from a mask that has been fit-tested and is worn correctly. EU classes use UK HSE ratings and US classes use OSHA’s. Cloth and surgical figures are typical values from fit studies; the ear-loop figure is an estimate from fit-test pass rates. Prices and weights are typical for the models listed, from makers’ and retailers’ specs.
In the animation, particles that get in are split between the filter and the edges to show the two paths; the total matches each mask’s rating.