OBJECTIVE To study the effect of an unpredictable drop in serum estradiol prior to hCG administration on pregnancy outcomes in in vitro fertilization cycles. METHODS 3653 consecutive IVF cycles from January 1, 1998 to December 31, 2000 at Brigham and Women's Hospital were reviewed, and 65 cycles in which oocyte retrieval (ER) was performed following a drop in serum estradiol (E(2)) not associated with intentional withdrawal of gonadotropins were identified. Daily gonadotropin dose was decreased at some time in 25 of these cycles, while the remaining 40 cycles did not have a reduction in gonadotropin dose. A retrospective case-control study of the respective live birth rates and pregnancy loss rates of patients with unpredictable E(2) drops in the 65 study cycles were compared to 65 age matched controls. RESULTS Live birth rates (32% vs. 35%, p=0.72) and pregnancy loss rates (28% vs. 30%, p=0.76) were similar for all study and control groups respectively. There were no differences in live birth and pregnancy loss rates in cycles undergoing gonadotropin dose reduction (40% vs. 44%, p=0.78 and 29% vs. 39%, p=0.70) and cycles without gonadotropin dose reduction (28% vs. 30%, p=0.81 and 27% vs. 20%, p=0.72). CONCLUSIONS In the absence of coasting, a drop in serum estradiol levels during GnRH-agonist downregulated controlled ovarian hyperstimulation for IVF prior to hCG is not associated with a decrease in live birth rates or pregnancy loss rates.