In previous studies, we reported a high rate of spontaneous abortions in insulin-dependent diabetic pregnancies. Abortions were associated with poor first-trimester glycemic control. We hypothesized that improvement of glycemic control from one pregnancy to the other would improve fetal outcome and that deterioration of glycemic control would increase the likelihood of abortion. We studied prospectively 43 insulin-dependent diabetic women (White class B-RF) with two consecutive pregnancies, recruited before 9 weeks' gestation. Preprandial and 90-minute postprandial blood glucose concentrations were measured at each weekly visit. Glycohemoglobin A1 was measured at 9 weeks' gestation. Twenty women had two successful pregnancies and 15 had an abortion followed by a successful pregnancy (abortion- no abortion); the sample sizes for other sequences (no abortion-abortion, N=5; and abortion-abortion, N=3) were too small to allow for analysis. Glycohemoglobin A1 concentrations were stable in the sequence no abortion-no abortion (9.7 ± 0.5 versus 9.8 ± 0.4%, mean ± SEM; not significant), whereas in the sequence abortion-no abortion, there was a significant decrease in glycohemoglobin A1 values from the nonsuccessful to the successful pregnancy (10.7 ± 0.6 versus 9.3 ± 0.4%; P=.01). Similarly, in the sequence abortion-no abortion, there was a significant decrease in mean postprandial blood glucose from first to second pregnancy (166 ± 13 versus 135 ± 11 mg/dL; P=.04), whereas in the sequence no abortion-no abortion, mean postprandial blood glucose did not change significantly (160 ± 14 versus 144 ± 11 mg/dL; not significant). Logistic regression analysis, taking into account variables such as maternal age, age at onset of diabetes, and White classification, confirmed a decrease of glycohemoglobin A1 at 9 weeks as the best indicator of a successful outcome after a spontaneous abortion. We conclude that improvement of glycemic control from one pregnancy complicated by abortion to the next one is associated with improved outcome in the second pregnancy. Because glycohemoglobin A1 at 9 weeks reflects early first-trimester glycemic control, we speculate that patient education and prenatal counseling which followed the first abortive event were successful in improving both glycemic control and fetal outcome.