One of the chief complexities of genetic influences on human obesity appears to be gene-gene interactions. Here, we employed model-free approaches to look for gene-gene interaction effects in human obesity using genome scan data from 260 European American families. We found consistent evidence for statistical interaction between 2p25-p24 (18-38 cM) and 13q13-q21 (26-47 cM). For discrete traits, the positive correlations were significant at P<0.0001 (P</=0.0023 after correction for multiple tests) in both IBD-based and NPL-based analyses for BMI>/=40 kg/m(2). Other analytic approaches gave consistent, supportive results. For quantitative traits, interaction effects were significant for BMI (P=0.0012), percent fat (P=0.0265) and waist circumference (P=0.0023) in a Haseman-Elston regression model, and for BMI (P=0.0043) in variance component analysis. Our findings suggest that obesity-susceptibility loci in chromosome regions 2p25-p24 and 13q13-21 may interact to influence extreme human obesity. The identification of gene-gene interactions may prove crucial to understanding the contributions of genes, which, by themselves, have relatively small effects on obesity susceptibility and resistance.