The prediction of clinical response in depression has been based primarily on clinical symptoms and history. Recently, psychobiologic measures have been used to increase the accuracy of clinical prediction. In 34 drug-free patients with primary endogenous depression treated with amitriptyline, the application of EEG-monitored sleep criteria alone was more significant than clinical status alone in the prediction of clinical response. Prolonged REM latency and reduced difficulty in sleep onset following the administration of amitriptyline were the main sleep variables contributing to this prediction equation. These data suggest a strong relationship between clinical outcome and psychobiologic profile in patients with endogenous depression after a "pharmacologic probe" with a tricyclic antidepressant.