A total of 20 cases of heterotopic pregnancy were encountered among 2650 clinical pregnancies (0.75%) resulting from in-vitro fertilization/embryo transfer at Bourn Hall Clinic (Cambridge, UK) during the period July 1984-July 1993. The aetiology of heterotopic pregnancy in the series is multifactorial, with tubal damage as the main factor. Transvaginal ultrasonography showed a high sensitivity for making correct diagnoses of heterotopic pregnancies compared with transabdominal ultrasonography (93.3 versus 50.0%). The mean plasma human chorionic gonadotrophin (HCG) concentration on day 13 after embryo transfer was similar to those of uncomplicated intrauterine pregnancies and hence was of no diagnostic value. The serial plasma HCG concentrations of patients who delivered were significantly higher than for those who aborted their intrauterine pregnancies (P < 0.01), although the sample of data available was too small to make firm inferences. It does appear that serial HCG concentrations may have a predictive value of fair accuracy regarding the outcome of the intrauterine pregnancy in heterotopic pregnancies. The clinical presentations of the 20 cases at first examination were quite variable, with 45% (9/20) of patients asymptomatic. Tubal pregnancy in one patient resolved spontaneously, two cases were treated by an injection of potassium chloride into the gestational sac and the remaining 17 cases were treated by salpingectomy. In 10 patients the intrauterine pregnancy resulted in live birth and the remaining 10 patients aborted spontaneously.