Regressive models were used to search for possible major gene effects on pulmonary function in two groups of families: one ascertained through patients with chronic obstructive pulmonary disease [COPD defined as forced expiratory volume in one second (FEV1) less than 70% forced vital capacity (FVC)] and the other ascertained through patients with non-pulmonary disorders. There were 85 COPD families with data on 270 individuals and 56 non-pulmonary families with data on 199 individuals. The analysis was done on residuals obtained from a regression of FEV1 on age, sex, race, height, and ascertainment group. Smoking status was incorporated directly as a covariate in the regressive models. Data on probands were excluded in this analysis as a partial correction for ascertainment bias. The best fitting model for the 85 COPD families included a major gene effect with sex specific variances, but no residual familial correlation. The best fitting model for the non-pulmonary families was one with no major gene effect and no residual familial correlation. Cigarette smoking was a significant covariate in both groups of families. Testing for heterogeneity showed a significant difference in the control of pulmonary function among these COPD and non-pulmonary families (X2 = 20.12 on 6 df; p = 0.0026). Major gene effects appear to be limited to these COPD families, while there was no evidence for major gene effects in the non-pulmonary families.