Although genes causing rare Mendelian forms of human obesity have provided much useful information about underlying causes of obesity, these genes do not explain significant proportions of common obesity. This review presents evidence that animal models can be used to uncover subtle genetic effects on obesity and can provide a powerful rigorous compliment to human association studies. We discuss the advantages of animal models of obesity, various approaches to discovering obesity genes, and the future of mapping and isolating naturally occurring alleles of obesity genes. We review evidence that it is important to map naturally occurring obesity genes using quantitative trait locus (QTL) mapping, instead of mutagenesis and knockout models because the latter do not allow study of interactions and because naturally occurring obesity alleles can interfere with cloning from mutagenesis projects. Because a substantial percentage of human obesity results from complex interactions, the underlying genes can only be identified by direct studies in humans, which are still very difficult, or by studies in mice that begin with QTL mapping. Finally, we emphasize that animal model studies can be used to prove that a specific gene, only associated with obesity in humans, can indeed be the underlying cause of obesity in mammals.