The extent to which given levels of caloric restriction will improve glycemic status but increase plasma ketone bodies in gestational diabetic women has received little attention. After reviewing the underlying physiology, we present data on two feeding studies investigating the question. In the first, a weight-maintaining approximately 2400-kcal/day diet was fed on a metabolic ward to 12 gestational diabetic women for 1 week. In the second week, subjects were randomized to a continuation of the 2400-kcal/day diet or to a 1200-kcal/day diet. Twenty-four-hour mean glucose levels remained unchanged in the control group but declined in the calorie-restricted group (6.7 mM or 121 mg/dl in week 1 vs 5.4 mM or 97.3 mg/dl in week 2) (p less than 0.01). Nine-hour overnight fasting plasma insulin also declined but oral glucose tolerance did not improve with caloric restriction. Fasting plasma beta-hydroxybutyrate rose in the calorie-restricted group, along with an increase in ketonuria, but not in the control group. A second study compared the impact of a 33% calorie-restricted diet or insulin to a full-calorie diet in a similar 2-week experimental design and measured hepatic glucose output and insulin sensitivity with dideuterated glucose before and during an insulin clamp. Diet in three subjects improved fasting and 24-hr mean glucose by 22 and 10%, respectively, whereas prophylactic insulin in three subjects produced 0 and 4% reductions, respectively. On average, ketonuria after a 9-hr fast declined to an equivalent degree with both treatments. Hepatic glucose output and insulin sensitivity were not statistically significantly altered by gestational diabetes or the therapeutic interventions compared to nondiabetic normal weight or obese pregnant controls. In conclusion, 50% caloric restriction improves glycemic status in obese women with gestational diabetes but is associated with an increase in ketonuria, which is of uncertain significance. An intermediate 33% level of caloric restriction (to 1600-1800 kcal daily) may be more appropriate in dietary management of obese woman with gestational diabetes mellitus and more effective than prophylactic insulin. Further studies are required to confirm these findings.