Intrauterine devices releasing 20 micrograms/day levonorgestrel were inserted in 10 women (ages 25-34). Bleeding and spotting patterns were recorded on a menstrual card during one year of follow-up. Blood samples were collected for radioimmunoassays of LH, estradiol (E2), progesterone (P) and levonorgestrel (LNG) and for sex hormone binding globulin (SHBG) 3 times a week during the 1st month of use, and twice a week during the 6th and 12th treatment months. Among the 10 women, two experienced irregular cycles with prolonged intermenstrual spotting, four had amenorrhea in the latter part of treatment months, while the other four had regular cycles. According to the serum levels of E2 and P, the hormone profiles were divided into four types of reaction: A) anovulatory, B) anovulatory but with high follicular activity, C) ovulatory but with luteal insufficiency, and D) ovulatory. Among the 29 treatment cycles, there were 10 D-type, 3 C-type, 13 B-type and 3 A-type of ovarian reactions: 44.8% of the cycles were ovulatory (C + D) and 55.2% were anovulatory (A + B). In general, serum levels of levonorgestrel were low in ovulatory cycles and were high in anovulatory cycles. The difference was statistically significant. There were marked individual differences. The decline of serum LNG from the 1st (492 pmol/l) to the 6th (320 pmol/l) treatment months was 34.9% on average. The amenorrheic cycles coincided mostly with the hormonal profile of ovulatory types, which indicated that the cause of amenorrhea is due to the local effect of levonorgestrel on the endometrium. The levonorgestrel levels were significantly correlated with serum SHBG, r = 0.8856, p less than 0.001, and with E2, r = 0.4661, p less than 0.05.