The concept of gestational diabetes goes back at least to 1946. Over the years, sophistication has increased regarding the pathophysiology of this condition. However, there is not universal agreement on how to diagnose gestational diabetes. The most widely used diagnostic criteria in the U.S. were validated by their predictive value for subsequent diabetes in the mother, rather than by their ability to identify risk to the fetus and newborn. The best available evidence supports the notion that the relationship between carbohydrate intolerance in pregnancy and adverse perinatal outcomes is a continuous one, and no single cutoff can separate pregnant women into those with high risk and those with no risk at all. Suggestions are made for arriving at appropriate, albeit arbitrary, diagnostic criteria.