Over 70 prevalence studies of schizophrenia published since 1948 are reviewed. Despite problems of method, the studies suggest that there may be significant differences in prevalence among different populations and that prevalence may change over time. Further research is needed on how much differences in diagnosis or the course of schizophrenia contribute to the variable prevalence rates. The paper suggests the following as areas for additional enquiry: a possible north-south gradient, special populations (e.g. American Hutterites, northern Swedes, western Irish, castes in India) and ongoing case registers in areas of suspected high and low prevalence.