Various studies have compared the efficacy of GnRH agonists (GnRH-a) and antagonists (GnRH-ant) for controlled ovarian stimulation (COS) in women undergoing IVF. Nevertheless, few data are available about the use of GnRH-ant in poor responders. Here, a flexible protocol providing a gradual increase in the dose of GnRH-ant in association with recombinant LH (rec-LH) administration is compared with the standard GnRH-a flare-up protocol in 133 women at risk for poor ovarian response. The mean number of metaphase 2 oocytes (primary end point) was significantly higher in the antagonist group (5.73 +/- 3.57 vs. 4.64 +/- 2.23, respectively; P<.05).