Patients with rapid-cycling bipolar disorder (RCBD) experience at least four episodes of affective illness (depression, hypomania, and/or mania) in a year and are frequently resistant to conventional treatments. Whereas the gender ratio among patients with bipolar illness is 1:1, bipolar women are three times more likely than bipolar men to develop a rapid-cycling course. Therefore, this illness warrants study because of its significant morbidity and because it can yield information about the effects of gender on the course of bipolar disorder. In addition, since rapid-cycling patients experience frequent mood shifts that are accompanied by dramatic changes in the sleep-wake cycle, they are an interesting population in which to study the relationships between sleep, circadian rhythms, and mood disorders. This project consists of three studies of the chronobiology of RCBD: (1) a cross-sectional study of patients in the depressed and hypomanic states; (2) an intensive longitudinal study of a small number of patients; and (3) the collection of naturalistic longitudinal data (mood ratings and sleep logs) on all patients in our RCBD clinic. In the cross-sectional study, patients are admitted to the hospital for brief, intensive chronobiological evaluations once shortly after they switch into hypomania, and once shortly after they switch into depression. The hospitalizations are each 48 hours in duration and consist of a 20-hour """"""""naturalistic day"""""""", during which patients sleep and eat ad lib, and a 28-hour constant routine. The constant routine is designed to minimize the effects of light, sleep, activity, and caloric loading on the patient's circadian rhythms. During the hospitalizations, patients wear rectal temperature probes and activity monitors, sleep is monitored by EEG, and blood is withdrawn every thirty minutes through an indwelling intravenous catheter. Blood samples are used to measure melatonin, TSH, cortisol, prolactin, and growth hormone. Data from the first ten patients indicate that there is a trend such that the onset of melatonin secretion and the maximum of the smoothed cortisol rhythm are phase-advanced (shifted earlier) in the hypomanic, compared to the depressed, phase. We are currently expanding the sample in this study to see if these findings reach statistical significance. In addition to this cross-sectional study, we are conducting intensive longitudinal studies of the circadian rhythms of melatonin, temperature, and sleep in several patients with RCBD. These studies allow us to understand in greater detail the relationship between these rhythms and the patients' mood states. Finally, we have accumulated a large naturalistic data set that has allowed us to demonstrate a circadian rhythm in the direction of mood switches in patients with RCBD, and to explore the relationship between prior sleep (sleep duration, times of sleep and wake onset) and mood.