Higher exposure to light at night, for example, owing to night shift work or decrease in sleep duration, may suppress melatonin production, which in turn may increase the reproductive hormone levels. High levels of steroid hormones, especially estrogens, may be associated with an increase of the breast cancer risk. This study investigated whether variation in the sleep duration during one entire menstrual cycle corresponds to estradiol saliva concentrations in healthy, urban women of reproductive age. During 2000-2001, 95 regularly menstruating women aged 24-36 in Poland collected daily saliva samples for one entire menstrual cycle. Saliva samples were analyzed for concentration of 17-beta estradiol (E2) using radio immunoassay. Information on the number of hours of sleep per night (sleep duration) was collected daily by questionnaires for one entire menstrual cycle. Using covariance analysis, after adjustments for sleep duration, we documented a positive relationship between the sleep variation (sleep coefficient of variation) and E2 levels in women of reproductive age. Mean levels of E2 differed significantly in women from the lowest sleep coefficient of variation quartile (13.93 pmol/l) in comparison with other quartiles (22.35 pmol/l), (P<0.001). The low sleep variation group, that is, the women who sleep regularly, had mean E2 levels 60% lower than other groups. These results suggest that sleep variation significantly correlates with E2 levels, whereas sleep duration does not show a statistically significant relationship. This study suggests that sleep variation may influence endogenous estrogens, which is of importance for risk of breast cancer.