OBJECTIVE: REM latency is usually 60-120 min, but under certain conditions, its latency may be less than 25 min, in which it is known as sleep onset REM period (SOREMP). In order to identify the factors responsible for the appearance of SOREMP, we used the nocturnal sleep interruption method to investigate whether REM propensity in normal nocturnal sleep, (i.e. circadian variations in REM sleep related to body temperature rhythm), could affect the rate of SOREMP appearance (SOREMP%). MATERIALS AND METHODS: After two adaptation and one baseline nights, we interrupted the nocturnal sleep, either in the second or the fourth cycle (early and late conditions, respectively) of 16 subjects and compared SOREMP% at the second sleep onset between these conditions by chi(2) test. Rectal temperature was measured. RESULTS: SOREMP% was found to be 58.1% in the early condition and 87.5% in the late condition - a significant difference. Body temperature dropped at the second sleep onset in both conditions but the drops did not differ significantly. CONCLUSION: We concluded that SOREMP% was affected by circadian variations in REM propensity but SOREMP% and body temperature drop did not show a linear relationship. Further studies to discriminate the influence of circadian rhythm factors and slow wave sleep on SOREMP% are called for.