OBJECTIVE: Little is known about the comparative risks and benefits of medical treatments for gestational diabetes mellitus (GDM). We conducted a systematic review of randomized controlled trials and observational studies of maternal and neonatal outcomes in women with GDM treated with oral diabetes agents compared with all types of insulin. DATA SOURCES: We searched four electronic databases from inception through January 2007. Terms for GDM, insulins, and oral hypoglycemic agents were used in the search. Two investigators independently reviewed titles and abstracts, performed data abstraction on full articles, and assessed study quality. METHOD OF STUDY SELECTION: Articles were excluded if they had no comparison group or did not use a standard diagnosis of GDM (3-hour, 100-g oral glucose tolerance test or 2-hour, 75-g oral glucose tolerance test). Nine studies met our inclusion criteria, four randomized controlled trials (n=1,229 participants) and five observational studies (n=831 participants). Data were abstracted on study characteristics, gestational age at treatment, medication dosage, and length of follow-up. Outcomes included glycemic control, infant birth weight, neonatal hypoglycemia, and congenital anomalies. TABULATION, INTEGRATION, AND RESULTS: Two trials compared insulin to glyburide; one trial compared insulin, glyburide, and acarbose; and one trial compared insulin to metformin. No significant differences were found in maternal glycemic control or cesarean delivery rates between the insulin and glyburide groups. A meta-analysis showed similar infant birth weights between women treated with glyburide and women treated with insulin (mean difference -93 g) (95% confidence interval -191 to 5 g). There was a higher proportion of infants with neonatal hypoglycemia in the insulin group (8.1%) compared with the metformin group (3.3%) (P=.008). The rate of congenital malformations did not differ between pregnancies treated with insulin and those treated with oral agents. Observational studies were limited by selection bias and confounding. CONCLUSION: No substantial maternal or neonatal outcome differences were found with the use of glyburide or metformin compared with use of insulin in women with GDM.