To determine the effect of long-acting intrauterine progestin on the impedance to blood flow in the uterine arteries, we applied a levonorgestrel-releasing (20 microg/day) intrauterine contraceptive system (LNG-IUS) to the uterine cavity of 27 fertile, regularly menstruating women. Measurements were performed using transvaginal colour Doppler ultrasonography in the mid-luteal phase and on day 1 of menstruation in the absence of the LNG-IUS and 3 months later in its presence. The mean (SD) mid-luteal uterine artery pulsatility index (PI) had risen from a pretreatment level of 2.28 (0.48) to 2.70 (0.67) after 3 months use of the LNG-IUS (P < 0.01), but as regards day 1 of menstruation, no change was observed. The increase in the mean PI (SD) at the mid-luteal phase was present only in subjects with serum levonorgestrel concentrations >200 pg/ml (n = 16) [pretreatment, 2.25 (0.47); with LNG-IUS, 2.83 (0.59), P < 0.001], and absent in those with serum levonorgestrel concentrations <200 pg/ml (n = 10) [pretreatment, 2.31 (0.53); with LNG-IUS, 2.51 (0.79), ns]. Mean progesterone concentrations were lower with the LNG-IUS in place in both groups. We conclude that a levonorgestrel-releasing intrauterine device appears to increase the impedance to blood flow in the uterine arteries during the mid-luteal phase in correlation with serum concentrations of LNG and a concomitant decrease in serum progesterone concentrations.