BACKGROUND: In ∼12–14% of young normogonadotrophic women treated with a depot GnRH agonist long protocol, the initial ovarian response to recombinant human FSH (rFSH) can be suboptimal. We have tested the hypothesis that these women may benefit from recombinant human LH (rLH) supplementation in a multicentre, prospective, randomized trial compared with patients treated with an rFSH step-up protocol. METHODS: A total of 260 young normogonadotrophic women undergoing controlled ovarian stimulation with a GnRH agonist long protocol for IVF/ICSI were enrolled. The starting dose of rFSH was 225 IU. One hundred and thirty patients with serum estradiol levels 5 mm but none >10 mm on both day 5 and day 8 of stimulation were randomly allocated to two groups. From the eighth day of stimulation, women in group A (n=65) received 150 IU of rLH in addition to rFSH, while those in group B (n=65) had an increase of 150 IU in the daily dose of rFSH (step-up protocol). One hundred and thirty normally responding women continued monotherapy with rFSH and served as a further control population (group C). RESULTS: The mean number of cumulus–oocyte complexes retrieved in group A (9.0±4.3) was significantly higher (P<0.01) compared with group B (rFSH 6.1±2.6) but significantly lower compared with group C (10.49±3.7, P<0.05). Implantation and pregnancy rates were significantly lower (P<0.05) in the rFSH step-up group (10.5 and 29.3% respectively) when compared with normal responders (18.1 and 47.3% respectively). CONCLUSIONS: rLH supplementation is more effective than increasing the dose of rFSH in terms of ovarian outcome in patients with an initial inadequate ovarian response to rFSH alone.