Obstetrical brachial plexus palsy remains an unfortunate consequence of difficult childbirth. Sixty-six such patients were reviewed. Included were 28 patients (42 percent) with upper plexus involvement and 38 (58 percent) with total plexopathy. The natural history of spontaneous recovery in all of these patients has been determined using an appropriate grading mechanism. Sixty-one patients (92 percent) recovered spontaneously and five patients (8 percent) required primary brachial plexus exploration and reconstruction (median age 12 months), demonstrating that most patients do well. Additional analysis was undertaken to examine ways in which outcome might be predicted. The analysis does not consider whether or not the patient was operated upon. Good or poor recovery was determined by the spontaneous recovery observed. Discriminant analysis revealed that whereas elbow flexion at 3 months correlated well with spontaneous recovery at 12 months, when used as a single parameter it incorrectly predicted recovery in 12.8 percent of cases. Shoulder abduction was not a significant predictor of recovery. Numerous other early parameters correlated well with spontaneous recovery. When elbow flexion and elbow, wrist, thumb, and finger extension at 3 months were combined into a test score, the proportion of patients whose recovery was incorrectly predicted was reduced to 5.2 percent.