The Taiwan Psychiatric Epidemiological Project, conducted from 1982 to 1986, used the multistage random sampling method with 5005, 3004 and 2995 subjects selected respectively from metropolitan Taipei (MT), 2 small towns (ST) and 6 rural villages (RV). The case identification tool was the Chinese modified Diagnostic Interview Schedule (DIS-CM). This study presents the lifetime and one-year prevalence of 27 and of 17 specific psychiatric disorders respectively. The lifetime prevalence of any disorder defined by the DIS-CM -- excluding tobacco dependence -- was 16.3%, 28.0% and 21.5% in the MT, ST and RV samples respectively. The differences in lifetime prevalence between the sexes and between the 3 sampling areas were significant for 15 and 8 disorders respectively. The ST sample seemed to have the most disorders, with the highest prevalence among 3 sampling areas. The mean ratio of one-year to lifetime prevalence was 0.67. The differences in prevalence rates between the 3 sampling areas and between the international studies are discussed from methodological, social and cultural points of view.