Saliva cortisol was measured at 11 p.m. in a sample of 74 psychiatric inpatients composed of 24 primary endogenous depressives, 40 secondary depressives and 20 nondepressives (DSM III and Saint-Louis criteria). Primary depressives had significantly higher 11 p.m. saliva cortisol levels than nondepressives (p < 0.02) and secondary depressives ( p< 0.05). In contrast, there were no significant differences between secondary depressive and nonde-pressive saliva cortisol levels. A saliva cortisol cutoff limit of 3.45 nmol/1 identified primary depressives with a sensitivity of 62.5% and with a specificity of 75% in the depressive group, and 90% in the nondepressive group. The measurement of saliva cortisol at 11 p.m. could be used alone as a reliable and practical index of hypothalamic-pituitary-adrenal axis activity in depression, especially in outpatients.