Recent studies suggest that gestational diabetes mellitus (GDM) is underdiagnosed. To test this hypothesis, we examined the relationship of perinatal complications to glucose tolerance during the third trimester. Our population consisted of 287 women evaluated at approximately 28 wk gestation who had normal fasting (less than 5.9 mM) and 2-h (less than 9.2 mM plasma glucose) levels after a 100-g glucose load. Glycosylated hemoglobin and glycosylated plasma protein were also measured. Study subjects were stratified into three groups based on 2-h plasma glucose values: group 1 (n = 59) less than 5.6 mM, group 2 (n = 112) 5.6-6.0 mM, and group 3 (n = 116) 6.7-9.2 mM. There were statistically significant but low correlations (r less than 0.20) between 2-h plasma glucose levels and mother's age, body mass index, infant weights, and Apgar scores. There was a significant increasing trend in the proportion of overweight and obese women from groups 1 to 3 (P less than 0.02). There was also a significant trend toward higher birth weights (P = 0.013) and larger proportions of large for gestational age (LGA) babies (P = 0.02) from groups 1 to 3, and women with LGA infants showed higher fasting and 2-h plasma glucose levels than women with non-LGA infants (P = 0.032). However, there was no significant difference in perinatal complications or infant morbidity or mortality between groups. Percentage of glycosylated hemoglobin or glycosylated plasma protein did not differ between groups.(ABSTRACT TRUNCATED AT 250 WORDS)