A group of 14 remitted patients, formerly hospitalized with unipolar depression, was matched individually with a group of normal controls. All subjects slept in the laboratory for five consecutive nights. Although remitted patients and controls slept for about the same length of time (6½ hours), remitted patients suffered from delayed sleep onset, showed more stage 1 and less delta sleep, and had a slower sleep cycle. Furthermore, night-by-night variability was much greater in remitted patients than in controls for practically all sleep measurements. These findings indicate that, in many patients, sleep is still disturbed more than six months after substantial clinical recovery from depression. Finally, variability among individuals was greater for remitted patients than for controls in some sleep measurements, suggesting less homogeneity among formerly depressed patients than is found among controls.