How Many Lives Do Smoke Alarms Really Save?

Last year, we put out a podcast called “Death By Fire? Probably Not.” It was about the remarkable decline in fatal fires in the U.S. over the past century, and explored some of the contributing factors.

Joseph M. Fleming, a deputy fire chief with the Boston Fire Department, has now written in with a guest post that challenges what we think we know about smoke alarms. Fleming has more than 30 years of experience in the fire industry (in both firefighting and management), and suggests that people think a little harder about smoke alarms.

Do Smoke Alarms Really Save Lives?

By Joseph M. Fleming

As a deputy chief on a major municipal fire department, I have preached for years about the life-saving benefits of working smoke alarms.  I also trust the lives of my family to the eight smoke alarms installed in my house.  However, some of the available data regarding smoke alarms raise disturbing questions about the actual effectiveness of smoke alarms at reducing fire deaths in the U.S.

In the illustrated edition of SuperFreakonomics, the following chart appears. It shows a continuous downward trend in fire deaths for the last 90 years and particularly for the last 50 years. 

Now I would like to provide information on how smoke alarm usage has increased over time.  Most of these smoke alarms were single-station, battery-powered ionization alarms.

Keeping this chart in mind, consider the following statement from the U.S. National Institute of Standards and Technology:

According to estimates by the National Fire Protection Association and the U.S. Fire Administration, U.S. home usage of smoke alarms rose from less than 10% in 1975 to at least 95% in 2000, while the number of home fire deaths was cut nearly in half. Thus the home smoke alarm is credited as the greatest success story in fire safety in the last part of the 20th  century, because it alone represented a highly effective fire safety technology with leverage on most of the fire death problem that went from only token usage to nearly universal usage in a remarkably short time.

If the ionization smoke alarm was responsible for most of the decrease in fire deaths in the last part of the 20th century, shouldn’t the rate of decrease have been greatest over the time period that smoke alarm usage increased the fastest?   Yet over the time period of 1977–1987, when the use of smoke alarms skyrocketed, the trend line remained relatively constant.  The death rate was trending down before smoke alarms and continued to trend down after they saturated the market.  It does not appear that ionization smoke alarms affected the trend line. NIST inexplicably ignores the trends in better building codes, reduction in smoking, better firefighting equipment, and better emergency medical care as likely reasons for the reduction in fire deaths.

I would like to analyze another statistic often cited to support the effectiveness of smoke alarms: “The death rate in fires with working smoke alarms (0.52 per 100 fires) was less than half (56% lower) than the risk of death from fires that did not have working smoke alarms (1.18 deaths per 100 fires), either because no smoke alarm was present or an alarm was present but did not operate.”

However, these numbers are skewed by including “confined fires” — i.e., fires that are contained to the object of origin.  Over this time period (2005–2009), no one died in this type of fire if the fire was big enough to operate the alarm.  If, using the same report, we only analyze “non-confined” fires, we get the following death rate per 100 fires for homes.

  • Smoke alarm present and operated – 1.15 (980 deaths / 85,100 fires)
  • No smoke alarm or alarm did not operate – 1.64 (1,640 deaths / 99,800 fires)

This is only a 29 percent reduction in death rate (1.15 versus 1.64). Given that some of the reduction is probably due to socioeconomic factors that accompany smoke alarm ownership, the reduction in risk attributable to the alarm is less than this percentage.  The numbers for apartments are even more troubling.

  • Smoke alarm present and operated – 1.17 (220 deaths / 18,800 fires)
  • No smoke alarm or alarm did not operate – 1.43 (200 deaths / 14,000 fires)

In apartments, smoke alarms only reduce the risk of dying in a fire by 18 percent (1.17 versus 1.42). 

It is highly probable that the main reason for the lack of effectiveness of operating smoke alarms is that most smoke alarms utilize ionization technology (the less-expensive kind of alarm). This technology has been shown to operate only after dangerous conditions have developed during smoldering fires, and these types of fires are extremely common during the times when occupants are sleeping and relying on the alarm to alert them.  It is one of the reasons that some states — Massachusetts and Vermont, for example — as well as the International Association of Firefighters recommend the use of photoelectric smoke alarms. (The eight smoke alarms in my house are photoelectric alarms.)  Another reason is that photoelectric alarms are far less likely to sound nuisance alarms and, as a consequence, are less likely to be disabled.

More information on this topic can be obtained at the following links:

Smoke Detector Technology and the Investigation of Fatal Fires

Smoke Alarm Information from the International Association of Firefighters


This still doesn't seem to address the core concern of having a smoke detector, "does this save lives of someone sleeping in the house when a fire breaks out?"

Your denominators (I assume), still contain house fires, and apartment fires, that aren't contained but still break out when the occupant is out of the house, or is aware of the fire, and is able to get out of the house.

In all of those cases, I would expect similar rates from "no detector" and "detector". But, what about the number of cases where the fire broke out while the occupant was sleeping?

All of a sudden, we might see 980 deaths/5000 fires and 1640 deaths/5000 fires. . .no longer such a small difference.


um, getting an ion alarm for a couple o bucks in order to decrease death by fire by 18% is a no-brainer- notwithstanding the annoyance to neighbors every time i burn the damn toast...

Aaron Fuller

Is it possible that the impact of smoke alarms has been blunted by compensating risky behaviour by users? I believe there have been some studies shown that as car braking technology improves, people's average speed - even subconsciously - goes up. So if, for example, people with smoke alarms were less likely to have taken other fire safety precautions, the smoke alarms MAY be as effective as advertised, but that their use should be caveated with statements like 'Alarms can improve fire response time, but fires themselves may be more likely.'

All that said though, I'd still in favour of them being encouraged, as long as any compensating risky behavioiur is less effective in raising deaths than the alarms are in reducing them. I can't see any 'downside' in their introduction and use, only that the upside may not be as dramatic as they seem. If (as was covered in another Freakonomics podcast in the context of 'drunk walking' being more dangerous per mile than drunk driving), their installation made people behave so foolishly as the total number of fires went up, then that would be a case for not installing alarms. But if the net effect is what's causing even a relatively low proporion of the overall decrease, I'd say go for it.



Most people don't even think about their smoke detectors when it is time to change the batteries, so I doubt they are increasing their risky behaviors due to them. Having been a firefighter for 20 years, I have found the lack of basic fire safety knowledge or common sense to be pervasive to the point of an epidemic. Unfortunately, we have lost so much funding we cannot maintain our inspection or public education programs. We have little time anymore to do anything but train for and respond to emergencies.


Know what else makes them less likely to be disabled? Allowing a temporary (say 8 hours) by-pass of the "low-batter warning."

I don't understand the logic of having the thing beep ever 5-10 minutes (the infrequency increases the time to locate which one's battery needs replaced) in the middle of the night. I wake up, find out which one it is (30 minutes later) only to realize I don't have any 9V batteries. I'll pick one up in the morning but why can't I hit a button to say, "I acknowledge the battery is low and needs replaced soon. I'll do so at my earliest convenience."

That feature would keep me from the only time I've disabled mine: I put the thing outside so I could sleep.


Ok, so given that fire alarms are a minor factor in reducing death, can you discuss some of the best precautions people can take to prevent fires in the first place? For example, making sure wiring is up to code, etc.

Joel Upchurch

If you look at the fire deaths, you will notice that most of decrease happened before smoke alarms were available. I suspect most of the decrease is from safer heating systems. A lot of the heating systems used open flames and exposed heating elements. I grew up in a house with a wood stove with no automatic controls and before that we had a open fire place. I had neighbors who burned down two houses because they left their oven open to heat the house in the winter.

I wonder how many of the houses that lack smoke detectors also have antiquated heating systems? When my house was built in 2003 it came with 7 smoke alarms without me asking.

Also there have been big improvements in emergency care. I suspect that many victims are resuscitated that would have died a few decades ago.


Everyone should consider a monitored or "supervised" smoke/heat detector. These are photoelectric detectors that are monitored along with your home security system. Your system does not have to armed for them to work, so no matter if you are away from the home or asleep, if the smoke detector goes off, the monitoring station will immediately dispatch the fire department. I have been 4000 miles away from home and received a call from my local fire department while they were at my home responding to an alert from my smoke detector -- that's peace of mind that is well worth the small monthly fee for a monitored smoke detector!! I'm convinced that hese types of detectors are the ones that save lives, pets, and property.


Guess this is evidence that some of us live in vastly different worlds. Home security system? Not unless you count a couple of dogs (who'd probably lick an intruder to death) and some friendly neighbors.


I suspect a lot of the decrease post 1965 or so is due to the decline in cigarette smoking. At least superficially, the declines seem to be similar: There are probably a bunch of linked secondary effects, too. For instance, how many fires are attributed to kids playing with matches? If the parents don't smoke, they're far less likely to leave matches &c where the kids can easily get them.


If you look at the major causes of house fires you will see that the great majority of fires will occur when the house occupants are awake; the situation in which a smoke alarm is least effective. More than half of all fires start in the kitchen when someone is cooking. Probably the only common fire source that is just as likely to occur when occupants are asleep is the heating system. Heating system fires are about 5% of all house fires.

Smoke alarms may very well be like air bags; useful only in a small proportion of cases or as a backup for other much more effective safety measures (seat belts). Unfortunately research in these areas tends to be done by those who are pre-disposed to find what is expected; either manufacturers, or the government offices that mandated the devices. The few that use careful data analysis to challenge the official wisdom often walk into a political firestorm because everyone "knows" they must be wrong.



As Mr. Fleming says the results probably aren't even as good as his adjusted numbers. I rent apartments and the city requires detectors in each BR and outside each BR, so I see a lotof smoke alarms. Guess which ones are disabled? the less educated / lower income (as mentioned) and also those who suffer lots of nuisance alarms. Who suffers nuisance alarms? Those who smoke cigarettes or pot. Smoking being of course a significant source of fires and one that I assume is somewhat likely to occur while sleeping and/or passed out.


It's not just bad behavior that is responsible for disabling smoke detectors.

I had to disable the smoke alarm at my apartment because it was installed in the kitchen and caused too many false alarms from normal cooking. I had one apartment where a hot shower could set it off. The hush button only last for 1 minute, which makes it a chore to silence.

The designers of smoke alarms focus everything on safety above everything else, without considering real world usage. It is no different than when car alarms first came out.


This is the first I heard of the two different alarm types, and the issues with the ionization type rings true with my family. The fire alarm goes off practically every day whenever we cook (and no, we are not burning the food). There's been times it went off within minutes of turning on the oven. We keep the battery in now in the name of safety, but for most of my life it had been disabled due to the annoyance.

The problem now is that even though it is now kept working, our reaction to it has been dulled. Much like car alarms, when I hear a fire alarm, my first reaction is not that there is danger; more likely it's just another false alarm.

Sal's Dad

In my family, the first reaction to the alarm is "the chicken is done!"

Of course, this provides a regular test, and the kids know what the alarm means, should it ever go off in the middle of the night!

Klaus Esteran

Ok if we accept the statistics don't add up, there is still the question if smoke alarms help me individually

If they reduce the risk of dying in an apartment fire by 18% I know my answer. I'd like my death risk reduced by 18%. I really like the idea, and I of course wouldn't mind the rate being higher.

I once read some information that the average time a smoke alarm gains you is one minute. I can't find the reference at the moment. One minute isn't much additional time, but if that one minute, or one additional second, helps me to reach the door or window before the smoke gets me, I'd pretty please have that one minute.

So, my smoke alarms will stay, especially considering the cost.


Klaus, if you like your death risk reduced by 18%, wouldn't you really like your death risk reduced by 80-90%. The IAFF endorses photoelectric alarms for a reason and a number of states are now starting to mandate photoelectric alarms. Most security systems will not even install ionization alarms anymore. Why? Here is why; they get disabled nearly 5 times more than photoelectric alarms because of the nuisance alarms problems (they were originally designed to detect gases not larger particles such as smoke) and ionization alarms sound more than 30-60 minutes slower in smoldering fires than photoelectric alarms. The type of fire that kills you when you are sleeping and your senses aren't as keen. THIS is why more than 50% of all fire deaths occur during the sleeping hours. Ponder this information before you go to sleep.

Adrian Butler

The International Association of Fire Fighters (IAFF) is the largest fire fighter's union in the world with over 300,000 members throughout Canada and the USA.
The Australasian Fire & Emergency Service Authorities Council (AFAC) is the peak representative body of all Australian and New Zealand Fire Brigades.
The officially documented position on smoke alarms, of both the IAFF and AFAC is that:
"Ionization smoke alarms may not operate in time to alert occupants early enough to escape from smoldering fires."
The latest edition of 'The Volunteer Fire Fighter' magazine from here in Australia carries a very informative report (including information from Boston's Chief Jay Felming) about this issue titled, 'Ionization Smoke Alarms are DEADLY'.
Please take just a few minutes and check it out:

Thank you.
Adrian Butler
The World Fire Safety Foundation
Wooloweyah, NSW, Australia


Jay Fleming

I would like to respond to some of the comments regarding my article.

1st - I greatly appreciate Freakonomics publishing this information. It is a breath of fresh air to have people who are willing to listen to opinions that challenge the majority. I would also like to thank everyone who took the time to read and comment on my article.

2nd - I am not trying to argue that smoke alarms, i.e. ionization smoke alarms, do not help reduce the risk of dying in a fire. I am trying to argue that they are not nearly as effective as people assume they are. So although it is definitely worth spending $7 - $10 for an ionization detectors that might provide an 18% reduction in risk, it is far more effective to spend $10-$12 for a photoelectric detector that could provide a 50% reduction in risk.

3rd - A good way to analyze the benefit of alarms while people are sleeping is to look at the statistics regarding fires started with smoking materials. The vast majority occur while people are sleeping. According to the NFPA the number of deaths per 100 fires (started by smoking materials) has increased since the introduction of ionization smoke alarms. This would appear to indicate that for the type of fire that occurs while people are sleeping, which should be the kind of fire where smoke alarms could provide the biggest benefit, it has not helped that much.

4th - It is important for people to maintain there smoke alarms but since my argument focused on operating alarms the battery maintenance issue is irrelevant to my argument.

5th –Several people mentioned other factors that might be responsible for the decrease. I mention several in the article. I discuss this possibility in much more detail in the link listed earlier.

6th – In addition to providing better protection from fires, photoelectric smoke alarms are also far less prone to nuisance alarms. Several studies have shown that they are disabled 4-5 times less often than ionization.

7th – The best thing that people can do to reduce their risk from fire is to install photoelectric smoke alarms. The vast majority of Americans have ionization.