In 2016, Yale began following 3,635 people aged 50+; 12 years later, those reading books regularly showed a 23% lower mortality hazard, with readers reaching an 80% survival point 23 months later

A good book may do more than occupy a quiet afternoon, but one of the most repeated claims about reading and longevity is more complicated than it first appears. A Yale analysis of 3,635 Americans aged 50 and older found that people who read books had a lower risk of dying during as much as 12 years of follow-up. Those who read more than 3.5 hours a week had a 23 percent lower mortality hazard than people who did not read books. Yet the study does not show that reading itself adds years to a person's life.

The research, published in Social Science & Medicine in 2016, drew on data from the U.S. Health and Retirement Study, a large national survey that follows adults over age 50. Avni Bavishi, Martin Slade and Becca Levy examined whether book reading at baseline was associated with survival over the years that followed. The researchers were interested in more than whether older adults read, because they also compared reading intensity and accounted for a wide range of social, economic and health differences. That made the study more informative than a simple comparison between people who happened to enjoy books and those who did not.

The reading habit was measured in just one week

The study began with a surprisingly small snapshot of participants' lives. People reported how many hours they had spent reading books during the previous week, while newspaper and magazine reading was recorded separately. The researchers then placed participants into three categories: people who read no books, those who read up to 3.5 hours a week, and those who read more than 3.5 hours. That single week became the study's measure of a person's book-reading behavior at the beginning of the analysis.


This is important because reading habits are rarely fixed. Someone who reported no book reading during one particular week might normally read several books a month, while another person could have reported several hours because they happened to be reading a long novel during the survey period. The study could not continuously measure what participants read or whether their habits changed over the following decade. Even with that limitation, the researchers had a substantial amount of follow-up data to work with, totaling 34,496 person-years of observation.

Participants were followed for an average of 9.49 years, with some tracked for as long as 12 years. Across the full sample, 27.4 percent of participants died during the follow-up period. The raw numbers appeared to favor book readers: 33 percent of people who did not read books died, compared with 27 percent of those who did. But that difference alone could not establish that books were responsible, because readers and nonreaders differed in other ways that might influence longevity.

The 23 percent figure came from an adjusted model

The more widely cited result came after the researchers accounted for many of those differences statistically. People who read up to 3.5 hours per week had a 17 percent lower mortality hazard than nonreaders. For participants who read more than 3.5 hours per week, the reduction was 23 percent. When the two book-reading groups were combined, readers had a 20 percent lower adjusted mortality hazard than nonreaders during the follow-up period.

The statistical model included age, sex, race, education, wealth, marital status and employment. It also considered self-rated health, eyesight, depression and several diagnosed medical conditions. The association remained when the researchers examined groups divided by factors such as sex, education, wealth and baseline health. That persistence is one reason the study attracted attention, because the result did not disappear after researchers accounted for many obvious differences between readers and nonreaders.

Still, an adjusted association remains an association. Statistical models can account for measured characteristics, but they cannot guarantee that every relevant difference between two groups has been captured. Readers may differ from nonreaders in social connection, daily routines, cognitive activity, access to healthcare, physical activity or other behaviors that were not perfectly measured. Those factors could influence mortality independently of book reading.

Where did the famous 23 months come from?

The study contained another number that became especially memorable. Researchers plotted survival curves showing how the proportion of participants still alive changed over time. On the unadjusted curves, 80 percent of non-book readers reached the survival threshold after about 85 months. Book readers reached the same point after approximately 108 months. The difference was 23 months.

That does not mean book readers gained 23 months of life by reading. The comparison came from raw, unadjusted survival curves, meaning it did not incorporate the extensive statistical adjustments used for the mortality-hazard analysis. Once the curves were adjusted, the difference at that survival point was about four months rather than 23 months. The distinction is easy to lose when the headline number is repeated without explaining where it came from.

The 23-month figure therefore describes a difference between two observed survival curves, not a proven biological effect of reading. It is a useful descriptive statistic from the original analysis, but it should not be translated into a promise that reading a certain number of hours each week will extend life by nearly two years. That interpretation would go beyond what the research actually tested.

Why might reading be associated with survival?

The study did not prove a biological mechanism, but there are plausible explanations for why regular book reading might correlate with healthier aging. Reading requires sustained attention, language processing and the integration of information over time. A novel, for example, asks readers to keep track of characters, events, relationships and changing circumstances while building meaning from written language.

That kind of mental engagement could potentially contribute to cognitive stimulation, although this particular study cannot establish that mechanism. Reading may also be part of a broader lifestyle pattern. Someone who regularly spends time with books may have routines that provide structure, lower exposure to certain stressors or encourage other forms of intellectually engaging activity. The important point is that the study measured reading behavior and subsequent mortality, not the biological pathway connecting the two.

The researchers also found that the association was not simply explained by socioeconomic status or education. Wealth and education were included in the analysis, along with several measures of health. Yet statistical adjustment cannot eliminate every form of confounding. A person's social environment, personality, lifelong habits and access to resources can be difficult to capture in a dataset, especially when the key behavior was measured at only one point in time.

Newspapers and magazines were not the same as books

Another detail often overlooked is that the study distinguished book reading from reading newspapers and magazines. That allowed the researchers to examine whether the association was specifically connected with books rather than simply spending time reading any printed material. The distinction matters because different forms of reading place different demands on attention and comprehension.

Books generally require longer periods of continuous engagement. A reader may spend hours following a single narrative or developing an understanding of a complex subject. Newspapers and magazines can involve shorter, more fragmented encounters with information. The study's findings do not prove that this difference explains the survival association, but it offers one reason researchers considered book reading separately rather than treating all reading as one behavior.

What the Yale study can and cannot tell us

The strongest conclusion is relatively simple. In this large group of older Americans, book reading was associated with lower mortality over long-term follow-up. The association remained after researchers adjusted for numerous demographic, socioeconomic and health characteristics. People reading more than 3.5 hours a week showed a 23 percent lower mortality hazard than nonreaders in the adjusted analysis.

The weaker conclusion is that reading causes people to live longer. The study was observational, not a randomized experiment. Participants were not assigned to read books for a decade while another group was instructed not to read. Because of that design, researchers could identify a relationship between reading and survival but could not determine whether reading itself produced the lower mortality risk.

That distinction does not make the finding meaningless. Observational studies are often valuable for identifying patterns that deserve further research, particularly when studying long-term outcomes such as aging and mortality. But the responsible takeaway is narrower than the library-poster version. Reading was linked with better survival in this Yale analysis, while the famous 23-month difference came from unadjusted survival curves and should not be treated as an extra two years of life caused by books.

For readers, that may actually be the more interesting finding. A quiet habit that looks ordinary on the surface was associated with a measurable difference in long-term survival, even after researchers accounted for many major differences between people. Science has not established that a novel is a prescription for longevity. What the evidence does show is that the relationship between reading, aging, health and survival is worth taking seriously, while leaving room for the harder question that the study could not answer: whether books change the course of aging, or whether the people drawn to books already carry some of the characteristics that help them live longer.