Over eleven million births were recorded in the database, and 17,334 had a documented brachial plexus injury in the total of three years, yielding a nationwide mean and standard error of incidence of neonatal brachial plexus palsy in the United States of at least 1.51 +/- 0.02 cases per 1000 live births. The incidence of this condition has shown a significant decrease over the years (p < 0.01). In the multivariate analysis, shoulder dystocia had a 100 times greater risk, an exceptionally large baby (>4.5 kg) had a fourteen times greater risk, and forceps delivery had a nine times greater risk for injury. Having a twin or multiple birth mates and delivery by cesarean section had a protective effect against the occurrence of neonatal brachial plexus palsy. Forty-six percent of all children with neonatal brachial plexus palsy had one or more known risk factors, and fifty-four percent had no known risk factors.