We prospectively studied 110 asymptomatic female infertility patients with serial serum measures of beta-human chorionic gonadotropin (hCG), estradiol (E2) and progesterone (P) to determine their sensitivity, specificity, predictive value and test efficiency, alone or in combination, for the prediction of pathologic gestations prior to five weeks after ovulation. Circulating levels of serum beta-hCG, E2 and P were measured at 48- or 72-hour intervals. Seventy-four patients (67%) had viable pregnancies, for which the abnormal changes in steroid levels were defined as: a beta-hCG rise of less than 66% in 48 hours or less than 120% in 72 hours, an E2 decline of greater than 15% in 48 hours or greater than 20% in 72 hours, or a P decline of greater than 25% in 48 hours or greater than 33% in 72 hours. Thirty-six women (33%) had pathologic pregnancies, which included ectopic pregnancies (8), spontaneous or missed abortions (7), blighted ova (anembryonic gestation, 20) and hydatidiform mole (1). For the detection of pathologic pregnancies in this asymptomatic infertility population, the sensitivity of beta-hCG, E2 and P, singly or in combination, ranged from 34% to 78%, and the test efficiency ranged from 68% to 88%. Beta-hCG alone provided the highest sensitivity (78%) and test efficiency (88%). When compared to measuring serial beta-hCG alone, serum E2 or P did not enhance the test efficiency and lowered the sensitivity for the detection of pathologic pregnancies in an asymptomatic infertility population.