PURPOSE: To find the incidence of ectopic pregnancy (EP) in patients who conceived in the IVF-ET program, and risk factors affecting the occurrence of EP. METHODS: We analyzed the effects of the indication for IVF, type of ovarian stimulation (hMG + hCG vs. GnRH + hMG + hCG), type of embryo transfer (transcervical intratubal, intrauterine in chest-knee position and intrauterine in lithotomy position) and number of embryos transferred on the occurrence of EP. EP was treated by laparotomy, prostaglandin E2 or laparoscopic surgery. RESULTS: In 7991 stimulated and 92 natural cycles treated in the Ljubljana IVF-ET program between May 1983 and June 1994 we achieved 1059 pregnancies, of which 44 were ectopic (incidence 4.1%), the main risk being tubal factor infertility. There were 38 (86.3%) tubal, 3 (6.8%) heterotopic, 1 (2.4%) ovarian, and 2 (4.5%) cornual EP. In two patients multiple tubal EP occurred (1 twin, 1 triplet). Forty-two patients (95.4%) had tubal factor infertility, 1 (2.3%) unexplained, and 1 (2.3%) patient had male factor. The incidence of EP in patients with tubal infertility was 5.4%, in patients with unexplained infertility 2.0% and in those with male factor 0.9%. There appeared to be no correlation between the two superovulatory methods. With transcervical intrauterine ET the incidence of EP was 0 of 5 clinical pregnancies (CP); with intrauterine in chest-knee position it was 26 (3.5%) of 738 CP; with intrauterine in lithotomy position it was 17 (5.4%) of 316 CP. The difference between the two types of intrauterine ET is not statistically significant. The incidence of EP did not correlate with the number of embryos transferred. The average initial values of beta hCG performed 17 days after ET were significantly lower in patients with EP than in those with normal singleton pregnancy (157 +/- 143 mIU/ml vs. 408 +/- 148 mIU/ml). CONCLUSIONS: EP can complicate the IVF procedure. The main risk factor is tubal infertility with or without previous tubal surgery. The low initial value of beta hCG has a strong predictive value in the diagnosis of EP.