The extent to which glucose tolerance can be improved by a carbohydrate-reduced diet in pregnant women with latent diabetes was investigated and, furthermore, whether this measure can reduce the incidence of high birth weight in the infants. 1. In a group of pregnant women who were thought to have some carbohydrate, metabolic disturbance on the basis of adipositas, glycosuria, hereditary taint or a history of infants with birth weights over 4000 g, the oral glucose tolerance test (GTT), showed a significant improvement under diet during pregnancy in comparison with a control group without diet. 2. Although an increase in the mean GTT after birth is to be expected--and was observed in the control group--, under diet the GTT values actually decreased post partum. 3. The extent of improvement in the GTT values depended on the degree of carbohydrate reduction in the diet. 4. In the years 1968 to 1976 all pregnant women with latent (gestational) diabetes showing high GTT values were submitted to a carbohydrate-reduced diet. This measure contributed to a reduction in the incidence of birth weights over 4000 g from 7.8% to 4.8%.