In a sequential crossover study of IVF conducted from 2002 to 2006, growth hormone (GH) supplementation was assessed in poor-prognosis patients, categorized on the basis of past failure to conceive (mean 3.05 cycles) due to low response to high-dose stimulation (<3 metaphase II oocytes) or poor-quality embryos. Pregnancy rates in both fresh and frozen transfer cycles and the total productivity rates (fresh and frozen pregnancies per egg collection) were compared. In all, 159 patients had 488 treatment cycles: 221 with GH and 241 without GH. These cycles were also compared with 1572 uncategorized cycles from the same period. GH co-treatment significantly improved the clinical pregnancy rate per fresh transfer (P<0.001) as well as per frozen-thawed embryo derived from GH cycles (P<0.05) creating a highly significant productivity rate (P<0.001). The effect was significant across all age groups, especially in younger patients, and was independent of stimulation modality or number of transfers. GH cycles resulted in significantly more babies delivered per transfer than non-GH cycles (20% versus 7%; P<0.001) although less than the uncategorized cycles (53%). The data uniquely show that the effect of GH is directed at oocyte and subsequent embryo quality.