We use recently released, nationally representative data from the National
Health interview Survey-Multiple Cause of Death linked file to model the as
sociation of religious attendance and sociodemographic, health, and behavio
ral correlates with overall and cause-specific mortality. Religious attenda
nce is associated with U.S. adult mortality in a graded fashion: People who
never attend exhibit 1.87 times the risk of death in the follow-up period
compared with people who attend more than once a week. This translates into
a seven-year difference in life expectancy at age 20 between those who nev
er attend and those who attend more than once a week. Health selectivity is
responsible for a portion of the religious attendance effect: People who d
o not attend church or religious services are also more likely to be unheal
thy and, consequently, to die. However religious attendance also works thro
ugh increased social ties and behavioral factors to decrease the risks of d
eath. And although the magnitude of the association between religious atten
dance and mortality varies by cause of death, the direction of the associat
ion is consistent across causes.