Best smoke alarms

How to choose

Cover the whole house first

A working alarm matters more than which alarm you buy. In national fire data from 2012 to 2016, homes with no working alarm had 12.3 deaths per 1,000 reported fires and homes where alarms operated had 5.7, a 54% lower risk.¹²

You need an alarm inside every bedroom, outside each separate sleeping area and on every level, including the basement.³ NIST, the federal standards lab, found in full-scale house tests that alarms in bedrooms added escape time.¹ The Consumer Product Safety Commission (CPSC) says every home should have at least one CO alarm, which our CO and combo alarm guides cover.²³

Choose photoelectric

Home alarms use a photoelectric sensor, an ionization sensor or both. Ionization reacts first to flaming fires and photoelectric reacts first to smoldering ones.¹ In a 2007 statement on older alarms, NIST put ionization's average lead on flaming fires at about 50 seconds and photoelectric's lead on smoldering fires at about 30 minutes.²⁴

Daily life settles it for most homes. In a randomized trial of 761 Washington households, 20% of ionization alarms weren't working after 9 months, against 5% of photoelectric alarms, and a disconnected or missing battery was the most common reason.² In NIST's 2024 kitchen tests, older ionization alarms sounded in 43% of cooking tests with the range hood off, and older photoelectric alarms sounded in 12%.²⁵

Experts don't fully agree. The International Association of Fire Fighters calls photoelectric the better all-around choice, while USFA recommends both types or a dual-sensor alarm.^{3 26} No study settles that. Photoelectric throughout has the best support, and you can add a separate ionization alarm far from the kitchen for faster response to flaming fires.^{1 2}

The box may not name the sensor. First Alert sells its SMI models as Precision Detection, and its spec sheets show an ionization chamber.^{6 7} Don't buy ionization-only alarms in Massachusetts, which requires a photoelectric sensor in replacement battery alarms, or in Vermont, whose code sheet calls for photoelectric-only alarms in new installations.^{16 27}

Stay wired, or buy sealed

If your home has wired alarms, keep them. The National Fire Protection Association (NFPA) found that hardwired alarms operated in 94% of fires large enough to set them off, and battery-only alarms in 81%. Missing or disconnected batteries caused 43% of failures and dead batteries caused 25%.¹²

Without wiring, a sealed 10-year battery removes one cause of failure, the battery that someone pulls or lets die. In one program using lithium-battery alarms, residents said they had disabled 57% of the 37 alarms that weren't working, most often for chirping or cooking.²⁸ When researchers revisited homes 5 to 7 years after sealed alarms went in, 39% of the alarms were missing.²⁹

Some states decide this for you. New York stopped sales of battery-only alarms with removable batteries in 2019, so the AA-powered 20SDR isn't an option there.^{9 17} Massachusetts requires a sealed battery when you replace a battery alarm.¹⁶ Ask your fire department about local rules.

Link them if you can

Interconnected, or linked, alarms all sound when one does. In a CPSC survey of household fires that weren't reported to fire departments, linked alarms sounded in 53% of fires and were the only alert in 26%. Alarms that weren't linked scored 27% and 8%.¹² The survey is about 20 years old, but pay for linking before apps and voice prompts.

If your home has hardwired First Alert or BRK alarms, stay with that brand on the circuit and read each model's spec sheet, because the Precision Detection line includes ionization alarms.^{6 7 14} Our combo alarm guide covers a wired First Alert model with a photoelectric sensor. If you need links and have no wiring, the SC5 is one route, and the combo guide covers another.

Hang it away from the stove

Distance from the stove is the cheapest fix for false alarms. NIST advises at least 6 feet and ideally 20 feet from the cooking area.³⁰ In its tests, the same alarm model sounded in 32% of tests on the main ceiling panels and in 83% to 93% of tests when mounted closer to the stove. Running the range hood on high cut the new alarms from 31% of tests to 8%.²⁵

NFPA 72, the national alarm code, applies only where your area has adopted it. The 2025 edition bars alarms within 10 feet of a fixed cooking appliance unless that leaves no place for one, and then allows listed alarms at 6 to 10 feet. Between 10 and 20 feet, alarms must be listed for nuisance resistance.³¹

Check the mark on the alarm

Look for a UL or ETL mark on the alarm itself, not only in an online listing. Reviewed withdrew its recommendation of the X-Sense SC07-MR in March 2026 after finding that it doesn't carry UL certification.³² CPSC has also recalled marketplace alarms sold on Amazon, including LShome in 2026 and CHZHVAN combos in 2024. If you own either, stop using it right away and contact the seller for a refund.^{33 34}

Don't pay extra for a higher edition number. Listings had to meet the 8th or 9th edition of UL 217 by June 30, 2024, and products must move to the 10th edition by November 17, 2027. All of these include a cooking test, and the 9th edition added a flaming fire test right after it.^{35 36 37}

Those tougher tests should help with fires, since none of 45 older models met the 8th edition level in a NIST study.^{38 39} Cooking is less settled. NIST found the new alarms as a group weren't clearly better at ignoring cooking smoke, so treat any maker's claim of fewer false alarms as unproven.²⁵

Think about who sleeps there

Children can sleep through a standard alarm. In a study of 24 children, 58% woke to a tone and 96% woke to a recording of their mother's voice.⁴⁰ Retail voice alarms use generic prompts, which that study didn't cover, so treat a voice model such as Kidde's 20SD10-V as a reasonable extra and not a proven fix.¹⁰ People with hearing loss should look at alarms with strobe lights or vibration.³

Test it, count it, replace it

Test every alarm at least monthly. Kidde's sheets ask for a weekly button test on these alarms and say self-testing doesn't replace it. Replace each alarm 10 years after the date on it, even if it still passes.^{3 4} Treat 10 years as a maximum, because Kidde's footnotes say its warranty isn't a battery performance claim.^{4 41} Count your alarms after a move or a renovation. If cooking sets one off, open a window and press the hush button. Never pull the battery.³