A total of 75,979 women who were delivered vaginally in the period 1970 to 1985 were stratified into diabetic and nondiabetic groups. Overall, the incidence of macrosomia (greater than or equal to 4000 gm) was 7.6% (5674/74390) in the nondiabetic group and 20.6% (328/1589) in the diabetic group. Patients were further subdivided by weight categories at 250 gm intervals. Eight percent of shoulder dystocia occurred in the diabetic group when fetal weight was greater than or equal to 4250 gm. In contrast, 20% of shoulder dystocia in the nondiabetic group could have been prevented by elective cesarean section when the fetal weight was greater than or equal to 4500 gm. Furthermore, logistic regression analysis demonstrated that birth weight, diabetes, and labor abnormalities were the principal contributors to shoulder dystocia. Elective cesarean section is strongly recommended for diabetics with fetal weights greater than or equal to 4250 gm, and trial of vaginal delivery for nondiabetic fetuses with weights greater than or equal to 4000 gm is recommended. In all cases the clinician must be watchful for labor abnormalities in macrosomic fetuses.