An ethological method was employed to validate the DSM-III subtyping of unipolar, nondelusional depression. The nonverbal behavior of 44 depressed outpatients was video-recorded during psychiatric interview. The DSM-III subtyping was not significantly associated with sex or level of education. Patients with major depression (with or without melancholia) were significantly older than those with dysthymic disorder. The Hamilton Rating Scale for Depression (HRSD) scores of the 3 diagnostic groups indicated a progressive increase in symptom severity across DSM-III subtypes (dysthymic disorder less than major depression without melancholia less than major depression with melancholia). Ethological assessment failed to find any evidence for the validity of the DSM-III subtyping of unipolar depression. Of the 8 behavioral categories analyzed in this study, none showed statistically significant differences between the 3 diagnostic groups. Our interpretation of these results is that, whereas the DSM-III subtyping primarily reflects illness severity, the ethological profile measures a dimension of depression largely independent from severity, as indicated by the lack of correlation between the HRSD score and the categories of nonverbal behavior.