The Humboldt Family Study was conducted in the town of Humboldt, Saskatchewan, in 1993. Familial correlations and segregation analyses of lung function were carried out in 799 individuals in 214 nuclear families that included 214 fathers, 214 mothers, and 371 children. Forced expiratory volume in 1 second (FEV1) and maximal mid-expiratory flow rate (MMFR) were first regressed on age, height, weight, and their quadratic and cubic terms as well as on smoking status in four groups separately (mothers, fathers, daughters, and sons), with terms significant at the 0.10 level being retained. Residual phenotypes were standardized within the four groups. Class D regressive models were used to perform familial correlations and segregation analyses. For both FEV1 and MMFR, father-mother correlations were not significantly different from zero, and mother-offspring, father-offspring, and sibling-sibling correlations showed no statistically significant difference from each other. Based on the "polygenic" models, the estimated intraclass correlation is 0.132 (+/- 0.035) for FEV1 and 0.171 (+/- 0.039) for MMFR, and the narrow-sense heritability is 0.264 for FEV1 and 0.342 for MMFR. Segregation analysis shows that the "mixed" model with both single locus and polygenic components had a better fit for FEV1 than single-locus or polygenic only models. However, the model which included a nontransmitted environmental factor [tau(AA) = tau(AB) = tau(BB) = qA] and polygenic loci had a better fit than the Mendelian model [tau(AA) = 1, tau(AB) = 1/2, tau(BB) = 0] [Akaike's information criterion (AIC) = 2219.47 vs. AIC = 2222.14]. For MMFR, the Mendelian "mixed" model gave a nonsignificant improvement in loge likelihood compared to the simple polygenic model. Comparison of the single-locus model and Mendelian "mixed" model shows no difference in fitting the data. This study suggests that FEV1 and MMFR are controlled by many loci with no major effects and/or common environmental factors.