This is the first empirical report suggesting a menstrual-linked artifact in determining rates of depression. We investigated the effects of perceived menstrual cycle phase, and premenstrual and postmenstrual timing in the cycle, on scores for a self-report measure of depression that is widely used in epidemiological studies, the Center for Epidemiologic Studies Depression Scale (CES-D). The first study used a cross-sectional design and the second used a follow-up design with repeated measures across the menstrual cycle. Both studies assessed a community-based sample of regularly menstruating women. In the second study, testing was also done at 2-week intervals in men for comparison purposes. Both studies supported the hypothesis of a significant menstrual cycle-related effect on CES-D scores. The menstrual-linked effect is greater than the test-retest variability observed in men, such that women more often tend to meet the cut-off criteria for depression on the CES-D premenstrually, as compared with postmenstrually. Thus it is possible that cyclic variations in ratings on the CES-D may elevate epidemiological estimates of depression, as defined by RDC criteria, for menstruating women. Further longitudinal assessment in randomly selected samples will be needed in order to clarify whether a menstrual-linked effect is substantial and specific enough to require correction, or documentation of cycle phase, in epidemiological studies of depression.