A total of 136 adult subjects who met the Research Diagnostic Criteria for major (unipolar) depression were assessed for intensity of depressive symptomatology (using the Hamilton Rating Scale for Depression and the Beck Depression Inventory) and for lethality of current suicide ideation (using the Scale for Suicide Ideation). In addition, they were administered a variety of questionnaires assessing cognitive variables presumed to mediate depression and suicidality. Multiple regression analyses indicated that depressive symptomatology was best predicted by Beck's Hopelessness Scale in combination with the Detachment factor of the Crandell Cognitions Inventory. In addition, suicidality was significantly predicted by the Selective Abstraction and Overgeneralization factors of Lefebvre's Cognitive Error Questionnaire, when the effects of the Beck Depression Inventory were partialed out. The practical implications of these findings for discriminating suicidal from nonsuicidal depressives are discussed.