The English-language literature was reviewed to examine the evidence base for strategies that have been used to prevent ovarian hyperstimulation syndrome (OHSS). Prediction of OHSS by pretreatment patient characteristics and ovarian response parameters is unreliable, with a significant number of OHSS cases occurring in patients not thought to be high risk, while the majority of ‘high-risk’ cycles do not result in OHSS. Alternatives to ovarian stimulation should always be considered, depending on the clinical situation. Monofollicular ovulation induction with a cautious step-up regime carries a lower risk of overstimulation than step-down regimes. In in vitro fertilization (IVF) cycles, a low starting dose of follicle-stimulating hormone (FSH) and the use of 5000 iu human chorionic gonadotrophin (hCG) for final follicular maturation may benefit patients at high risk of OHSS. The role of GnRH antagonists is unclear. In women with polycystic ovaries, who are undergoing ovarian stimulation for IVF, metformin co-treatment may reduce the risk of OHSS. Coasting of cycles with over-response is associated with a reduced risk of OHSS, although precise criteria for initiating and ending coasting are not definable at present. Elective cryopreservation of all embryos prevents late OHSS, but its value has been poorly researched. The literature does not support a role for intravenous albumin, administered around the time of oocyte retrieval, in preventing OHSS. Evidence is insufficient regarding a possible role for hexa-ethyl starch. hCG should not be used for luteal support, as it is associated with a higher risk of OHSS, and equivalent pregnancy rates are obtained with the use of progesterone.