Non-insulin-dependent diabetes mellitus (NIDDM) has a high prevalence in Pima Indians. The disorder is familial, but the extent to which genetic factors are involved in its etiology is largely unknown. Segregation analysis was used to determine whether familial aggregation of NIDDM in this population could reflect the action of a single major gene. The analysis included 2,697 subjects from 653 nuclear families in which both parents and at least one offspring had been examined in the course of a longitudinal epidemiological study. The REGTL program of the SAGE package was used to fit models in which age at onset of NIDDM is transmitted from parent to offspring under the unified model for segregation analysis. Likelihood-ratio tests were used to test hypotheses related to genetic transmission. The hypothesis of no major effect was strongly rejected (P < .01), as was that of no transmission of the major effect (P < .01). Mendelian transmission was not rejected (P = .91). Similar results were obtained when covariates for obesity and birth cohort were added to the models and when a power transformation of age at onset was estimated. A strong effect of birth cohort with earlier age at onset in the later born cohorts was observed (P < .01). The findings are consistent with the hypothesis that a major gene influences the risk for NIDDM in Pima Indians by affecting age at onset. The expression of this gene may depend on environmental factors that have become more prevalent in recent-birth cohorts.