The equal-environment assumption (EEA), upon which twin methodology is based, was examined for the impact of physical similarity on phenotypic resemblance in five common psychiatric disorders: major depression, generalized anxiety disorder, phobia, alcoholism, and bulimia. A population-based sample of 882 female-female twin pairs of known zygosity was rated for similarity of appearance by color photographs. Psychiatric diagnoses were made by clinical assessment of personal interviews of the twins. Structural equation modeling of the data using physical similarity as a form of specified common environment provided no evidence for a significant effect of physical resemblance on concordance for major depression, generalized anxiety disorder, phobia, and alcoholism, thereby supporting the validity of the EEA in twin studies of these disorders. Results for bulimia, on the other hand, suggest, within the limitations of this study, that physical similarity may significantly influence twin resemblance for this disorder.