Objective. Intense debate exists concerning the number of homeless peo
ple in the United States. Previous studies, counting currently homeles
s people, have provided point-prevalence estimates of homelessness but
have been criticized on methodological grounds. This study reports li
fetime and 5-year prevalence estimates of, homelessness using a differ
ent methodological approach. Methods. Random-digit dialing was used to
interview 1507 adults living in households with telephones in the 48
contiguous States in the fall of 1990. Respondents were asked whether
they had ever been homeless:and if so, where they had slept while home
less. Results. Lifetime and 5-year prevalence of all types of homeless
ness combined were 14.0% (26 million people) and 4.6% (8.5 million peo
ple), respectively. Lifetime ''literal homelessness'' (sleeping in she
lters, abandoned buildings, bus and train Stations, etc.) was 7.4% (13
.5 million people). Five-year (1985 through 1990) prevalence of self-r
eported homelessness among those who had ever been literally homeless
was 3.1% (5.7 million people). Conclusions. The magnitude of the probl
em of homelessness is much greater than most previous attempts to enum
erate homeless people have led us to believe. This finding requires re
consideration of inferences about the causes of homelessness that were
derived from point-prevalence studies of currently homeless people.