Genetic predisposition contributes to scoliosis in humans. Two syndromes of primary scoliosis occur--congenital scoliosis, which presents at birth, often associated with other abnormalities, and idiopathic scoliosis which becomes apparent between infancy and adolescence. Little is known regarding the genetic transmission of scoliosis risk. Data gleaned from mouse mutations provide a valuable supplement to human family studies. More than 50 mouse mutations include scoliosis, kyphosis, or tail kinks as a phenotype; the locations of the human homologues for 28 of these can be predicted on the basis of synteny conservation. Some mouse mutations are either more penetrant or more fully expressed in one sex. The mouse data provide a basis both for optimism and for caution in understanding human scoliosis. Mouse models provide insight into mechanisms underlying spinal curvature and help direct searches for genes important in human disease. Four types of defects account for most mouse scoliosis: defects of cell-cell communication, intracellular signal transduction, matrix protein synthesis, and matrix protein metabolism. Mouse data suggest that at least two types of heterogeneity complicate genetic analysis: locus heterogeneity, in which lesions of distinct genes lead to a similar phenotype, and allelic heterogeneity, in which the phenotypes arising from alleles of a single gene differ. By focusing initial studies on multiplex families with apparent simple Mendelian inheritance the effect of heterogeneity is minimized.