Body dysmorphic disorder (BDD), together with its subtype muscle dysmorphia (MD), has been relocated from the Somatoform Disorders category in the DSM-IV to the newly created Obsessive-Compulsive and Related Disorders category in the DSM-5. Both categorizations have been criticized, and an empirically derived classification of BDD is lacking. A community sample of N = 736 participants completed an online survey assessing different psychopathologies. Using a structural equation modeling approach, six theoretically derived models, which differed in their allocation of BDD symptoms to various factors (i.e. general psychopathology, somatoform, obsessive-compulsive and related disorders, affective, body image, and BDD model) were tested in the full sample and in a restricted sample (n = 465) which indicated primary concerns other than shape and weight. Furthermore, measurement invariance across gender was examined. Of the six models, only the body image model showed a good fit (CFI = 0.972, RMSEA = 0.049, SRMR = 0.027, TLI = 0.959), and yielded better AIC and BIC indices than the competing models. Analyses in the restricted sample replicated these findings. Analyses of measurement invariance of the body image model showed partial metric invariance across gender. The findings suggest that a body image model provides the best fit for the classification of BDD and MD. This is in line with previous studies showing strong similarities between eating disorders and BDD, including MD. Measurement invariance across gender indicates a comparable presentation and comorbid structure of BDD in males and females, which also corresponds to the equal prevalence rates of BDD across gender.