Summary: The McCarley-Hobson model, describing the alternation of NREM and REM sleep in the cat, was applied to human electroencephalographic data. The influence of initial conditions on oscillatory behavior was especially emphasized. It appears that the distribution of REM latency in depression, the abnormal accumulation of REM sleep, the variability of NREM-REM cycle duration, the frequent stage shifts, and frequent aWakenings can be explained in this model by means of a decrease in the initial value of a single variable, which may be regarded as representing the strength of REM inhibition. The observation of slow wave sleep deficiency in depression may well be another reflection of this parameter. Key Words: REM sleep---Depression-Slow wave sleep---Ultradian rhythm.Numerous polygraphic studies have revealed that the sleep profile of endogenously depressed subjects is characterized by a shortened latency of rapid eye movement sleep (REMS), a relative overproduction of REMS in the first third of the sleep period, a reduction of slow wave sleep (SWS, i.e., stage 3 and 4), frequent stage shifts, frequent intermittent awakenings, and a higher incidence of REMs, especially in the first hours of sleep (1). The shorter REM latency, in particular, is considered to represent a specific feature of primary affective disorders (2,3). Vogel et al. (4) suggested that the abnormalities in REMS production in depression are caused by disturbances in an ultradian oscillator that regulates the NREM-REM alternations. The ultradian oscillator was assumed to be of the same nature as the one described for NREM-REM sleep regulation in the cat by McCarley and Hobson (5) and Hobson et a1. (6). In this model the oscillatory capacity is thought to be achieved by the reciprocal interaction of two elements, one excitatory and active during REMS, the other inhibitory and predominantly active during NREMS (Fig. 1). Vogel et al. (4) postulated that three aspects of this NREM-REM oscillator system are disturbed in depression: the strength of self-excitation of the REM-active part, the strength of the inhibition by the NREM-active part, and the power of the NREM-active part at sleep onset. The McCarleyHobson model, however, provides a different interpretation of sleep abnormalities in depres-