Trauma that occurs as a result of shoulder dystocia is an important cause of neonatal morbidity. If the occurrence of severe shoulder dystocia, resulting in fetal asphyxia and trauma, could be accurately predicted from maternal risk factors, then a cesarean section would be indicated to prevent the poor outcome. The information available in the obstetric literature, however, is contradictory regarding whether shoulder dystocia can be predicted. In the present study, the patients at greatest risk of shoulder dystocia (all 394 mothers delivering neonates with birth weights greater than or equal to 4000 gm over a 2-year period) were examined. A three-way discriminant analysis was used to determine if a model could be developed that could effectively predict those patients who would be included in each of the groups of no shoulder dystocia, shoulder dystocia without trauma (29 patients), and shoulder dystocia with trauma (20 patients). Three factors, including birth weight, prolonged deceleration phase, and length of second stage labor, were found individually to contribute significantly to the classification. However, when examined in detail, it was noted that while 94% of cases with no shoulder dystocia would be detected, only 16% of the cases of shoulder dystocia with trauma would be predicted by this model. We conclude that in the group of pregnancies delivering neonates greater than or equal to 4000 gm, the occurrence of shoulder dystocia cannot be predicted from clinical characteristics or labor abnormalities, and that the occurrence of shoulder dystocia is not evidence of medical malpractice.