A retrospective sample of 164 pregnancies in the years 1956 to 1975 of 78 insulin-dependent diabetic women was examined to assess the extent of the risk of clinically recognizable spontaneous abortion relative to the expected risk in a comparable population. This study was done before the implementation of the policy to establish good control of maternal diabetes before conception was attempted. After control of data for pregnancy number, maternal age, and epoch, the risk of spontaneous abortion among the diabetic women was almost double the estimate of the expected risk (p less than 0.001).