The effect on fetal and maternal outcome in gestational exercise in a recreational athlete was investigated in a 2 1/2-year prospective study. A training effect in a pregnant subject engaged in aerobic exercise was also determined. One hundred forty-one low-risk pregnant subjects had their aerobic capacity tested during the first trimester, during the third trimester, and post partum. Predicted maximum oxygen consumption values were determined by using standard Bruce treadmill protocol during the first and third tests and the Astrand protocol for submaximal testing on a bicycle ergometer during the third trimester. Exercise prescriptions were given. The primiparous and multiparous subjects were randomized into control and exercise groups, differing only in the frequency of aerobic exercise. The results showed that exercise had no associated increase in neonatal morbidity and obstetric complications. All subjects improved or maintained their level of aerobic fitness. The exercisers showed a significant training effect. The disparity in the second test's results may suggest that the physiologic characteristics of pregnancy masked the aerobic improvement.