We performed a prospective randomized study of goserelin, a long-acting gonadotrophin-releasing hormone agonist (GnRHa) and human menopausal gonadotrophin (HMG) versus clomiphene citrate and HMG for follicular stimulation in assisted reproduction to investigate whether the use of this GnRHa provides a clear advantage in terms of pregnancy per treatment cycle in unselected patients, who entered a first trial of assisted reproduction. From a retrospective analysis comparing the two stimulation protocols, a relative increase of the pregnancy rate per cycle of 50% was anticipated. To detect this difference with a power of 90%, 300 patients had to be included. The main prognostic factors affecting the outcome of assisted reproduction were equally divided among the two groups by a minimization procedure. The pregnancy rates per cycle were significantly better in the goserelin/HMG group than in the clomiphene citrate/HMG group, both for all procedures of assisted reproduction combined (36.8 versus 24.5%; P < 0.02) and for the main procedure of in-vitro fertilization (IVF) (37.0 versus 23.5%; P < 0.02). Differences in pregnancy rates per oocyte retrieval and per embryo transfer were less pronounced (37.8 versus 30.8%; P = 0.40 and 44.4 versus 36.8%; not significant). On the other hand, stimulation with goserelin/HMG was associated with a higher number of ampoules of HMG (44.9 versus 9.9; P < 0.0001), a longer duration of stimulation (11.2 versus 8.7 days; P < 0.0001) and an incidence of ovarian stimulation of 4.5% (7/154) versus 0% in the clomiphene citrate/HMG group. Goserelin was well tolerated and proved to be very reliable as an adjunct of follicular stimulation in assisted reproduction.(ABSTRACT TRUNCATED AT 250 WORDS)