A growing body of literature suggests that retrospective recall of psychiatric symptoms is often inaccurate and may distort knowledge about the course of illness and impact treatment. The current study examined the accuracy of retrospective recall of a variety of depressive symptoms in hospitalized depressed patients and nonclinical controls. Using the Experience Sampling Method, we compared average momentary symptom reports of 1 week to retrospective summaries of the same period. The depressed group exhibited negative biases in their recall of experienced anhedonia, sadness, confusion, and suicidality, but were relatively accurate in recall of helplessness, detachment, and self-control. Controls exhibited a different pattern; they were relatively accurate in their retrospective recall of confusion, suicidality and sadness, but exhibited positive biases in recall of anhedonia, helplessness, detachment, and self-control. Both groups exhibited comparable negative biases in their recall of experienced tension, difficulty concentrating, guilt, and fear. The findings suggest that for maximum accuracy in the assessment of depressive symptoms, scientists and practitioners should supplement retrospective self-reports with momentary measures, and consider using ambulatory assessment in cognitive behavioral treatments of depression.