This study examined the reliability and validity of a two-factor structure of the Center for Epidemiologic Studies Depression (CES-D) scale. The study was conducted in a large group of cancer patients (n = 475) and a matched reference group (n = 255). Both groups filled in a questionnaire at two points in time; patients 3 and 15 months after diagnosis. Factor analysis confirmed our hypothesis that the 16 negatively and four positively formulated items measure two relatively independent factors, i.e. Depressed Affect and Positive Affect. Therefore, these items should not be combined into an overall sumscore. In both groups, Depressed Affect proved to be a reliable and valid measure of depressive symptomatology, as indicated by its good internal consistency, its strong correlations with other measures of psychological distress and neuroticism, and its effectiveness in discriminating patients from the reference group on depressive symptomatology. In contrast, the validity of the Positive Affect factor could not be confirmed, since it was only weakly related to other measures of psychological distress and extraversion. Depressed and Positive Affect were about equally related to self-esteem, life satisfaction, and quality of life. These findings support the use of a sumscore based on the 16 negatively formulated CES-D items as a more valid measure of depressive symptomatology, in cancer patients and in healthy individuals from the general population.