Preventive medical interventions should be based on the highest level of scientific evidence. Actual criteria for diagnosing gestational diabetes mellitus (GDM) are neither uniform nor based on pregnancy outcomes. An expert panel from the International Association of Diabetes in Pregnancy Study Groups recently proposed that all pregnant women undergo a one-step 75 g OGTT, and defined new lower cut-off points to diagnose GDM (Metzger BE et al. Diabetes Care 33: 676-682). These criteria will double the prevalence of GDM, as 18% of all pregnant women will be labelled as abnormal. A recent article in Diabetologia (Ryan EA 54:480-486) claimed that maternal glucose is a weak predictor of big babies, that a single OGTT is poorly reproducible, and that expected benefits from intervention would be, at best, modest. This Commentary discusses other objections and argues that guidelines on any new GDM diagnostic strategy should be based on the results of randomised controlled trials rather than on disputable expert opinions.