Self-report data on smoking initiation (whether the respondent admitted ever having smoked) were obtained from three large adult twin samples (Australia, N = 3,808 pairs; Virginia, N = 2,145 pairs; AARP, N = 3,620 pairs). Data were broken down into birth cohorts, and genetic models were fitted to test whether the decline, in more recent birth cohorts, in the percentage of individuals becoming smokers has led to a change in the relative contributions of genes and environment to risk of becoming a smoker. Despite a marked change in the proportion of male respondents who reported ever having smoked, we found no evidence for cohort differences in genetic and environmental effects (no Genotype x Cohort interaction). Significant differences in genetic and environmental parameters were found between sexes, and between the Australian and the two U.S. samples. In the U.S. samples, estimates of the genetic contribution to risk of becoming a smoker were 60% in men, 51% in women. In the Australian sample, heritability estimates were 33% in men, but 67% in women. Significant shared environmental effects on smoking initiation also were found, accounting for 23% of the variance in U.S. men, 28% of the variance in U.S. women, 39% of the variance in Australian men, and 15% of the variance in Australian women. In models that allowed for the environmental impact of cotwin smoking on a twin's risk of smoking initiation, estimates of the direct genetic contribution to risk of smoking initiation were comparable or higher (49-58% in U.S. women and 71% in Australian women; 58-61% in U.S. men, and 37% in Australian men).