Methods of delivery, maternal and neonatal characteristics were examined to determine their role in the occurrence of Erb/Duchenne's palsy. Data from 210,947 Washington state birth certificates from 1980 through 1982 were examined. The incidence was 50.2 cases of Erb's palsy per 100,000 live births. A case control study design was used to analyze 106 cases and 386 controls by both univariable and multivariable analysis. Birth weight was shown to be a significant risk factor regardless of which method of delivery was used. A high birth weight infant (4001-4500 g) had 2.5 times the risk of incurring an upper brachial plexus injury compared with normal size infants (2501-4000 g). The risk for infants greater than 4500 g increased another tenfold (OR = 21.0). When birth weight was controlled for in the analysis, midforceps (OR = 18.3), vacuum extraction (OR = 17.2), and low forceps (OR = 3.7) remained significantly associated with the Erb's palsy. Delivery by cesarean section was associated with a significant protective effect (OR = 0.5) compared with instrumental vaginal delivery. These data demonstrate a high risk for serious birth injury associated with instrumental midpelvic delivery.