OBJECTIVE: To investigate a major gene hypothesis for body mass index (BMI) in a large sample of probands (n = 2580, ages 37-57 years) who were selected for obesity (BMI> or =34 kg/m2 for males and > or =38 kg/m2 for females), along with their spouses and first-degree relatives (n = 11,204 family members). The probands were recruited as part of an intervention trial assessing whether mortality and morbidity were improved after surgical intervention for obesity as part of the Swedish Obese Subjects (SOS) study. METHODS AND PROCEDURES: The current analyses were based on BMI measures obtained before intervention. Segregation analysis was carried out using the mixed model implementation in PAP (Pedigree Analysis Package), which allowed for ascertainment correction and for genotype-dependent effects of covariates (sex and age) in both the major gene component and the multifactorial (i.e., polygenic and familial environment) component. RESULTS: Both a major effect and a multifactorial effect were significant. The percentage of the total variance accounted for by the multifactorial effect was 17%-24% (increasing as a function of age), and by the major effect, 8%-34% (decreasing as a function of age). Although tests on the transmission probabilities (taus) were not compatible with Mendelian expectations of 1, 1/2, and 0, the equal taus model was rejected (i.e., the effect is transmitted in families) and the point estimates (0.96, 0.60, and 0.17) compared favorably to Mendelian expectations. The major effect was transmitted in a codominant fashion, consistent with a gene-environment interaction. DISCUSSION: These results suggest both multifactorial and major effect etiologies for BMI in these families of extremely obese probands. Before 20 years of age, the major effect dominates the BMI expression, but after age 20, multifactorial effects account for the most variance. Although the major effect is transmitted in these families, the pattern does not appear to be consistent with a simple Mendelian trait. The possibility of additional major loci (i.e., epistasis) and gene by environment interactions may explain these findings.