Objective: To examine the effectiveness of a partially home-based, moderate-intensity aerobic exercise program for women with gestational diabetes. Methods: This was a randomized experimental design. Thirty-three women with gestational diabetes were randomly assigned to the exercise or the no-exercise group. Subjects underwent hemoglobin A1C assay and submaximal cycle ergometer fitness tests at baseline and at study conclusion. Subjects kept diaries of home fasting and 2-hour postprandial blood glucose determinations. Exercise subjects were asked to exercise for 30 minutes three to four times weekly at 70% of estimated maximal heart rate for the weeks of study participation. Two exercise sessions weekly were supervised by the investigator, and two were unsupervised at home. Control-group subjects were asked to maintain their current activity level. Results: Daily fasting and postprandial blood glucose levels, hemoglobin A1C, incidence of exogenous insulin therapy, and incidence of newborn hypoglycemia were not different between the groups. There was a training effect in the exercise group (P = .005) but not in the control group (P = .25). A significant decline in daily grams of carbohydrate consumed was observed in the control group (P = .03), but not in the exercise group (P = .97). No complications were found in the subjects who exercised. Conclusions: A partially home-based exercise program did not reduce blood glucose levels, but did result in a modest increase in cardiorespiratory fitness. The intervention appeared safe.