Epidemiologic studies of suicide in the military have not controlled for th
e higher suicide rates of the unemployed expected in comparative national p
opulations. This study compared the observed number of suicides among U.S.
Marine Corps personnel from 1990 to 1996 with the expected number based on
rates for the employed general U.S. population. Standardized mortality rati
os were calculated to identify demographic groups with higher or lower than
expected numbers of suicides. The scan statistic and the Knox technique we
re used to evaluate potential suicide cluster patterns. Overall, there were
fewer suicides in the Marine Corps (n = 213) than expected (n = 225). Hisp
anic and other ethnic group males and female Marines had greater than expec
ted numbers. Evidence for suicide clustering in time and space was equivoca
l.