Secular changes in the rates of mania have been reported in the relatives of affectively ill patients, and the present study is an independent replication of these findings in a large probability sample of a community. Data from the five U.S. sites of the National Institute of Mental Health's Epidemiologic Catchment Area study (ECA), a community survey of adults, are presented. A total of 17,827 respondents were divided into eight birth cohorts, each spanning a 10-year interval. Actuarial life-table analyses showed a greater cumulative risk of mania for white men and women from the three most recently born cohorts (i.e., those born after 1935). The Wilcoxon test for homogeneity indicated an overall significant difference by birth cohort. These findings in a community sample are consistent with those previously reported in relatives of affectively ill patients. The results support the hypothesis that birth cohort changes have occurred in the risk of mania and that the risk was greatest in the post-1935 cohorts.