To identify variables that increase the chance of neonatal scalp injury during vacuum extraction. We conducted a prospective observational study of 134 vacuum extraction-assisted deliveries at Olive View-UCLA Medical Center in 1995. Data collected included parity, gestational age, duration of first and second stages of labor, indication for operative delivery, station and position of fetal head, duration of vacuum application, number of “pop-offs,” neonatal weight, and descriptions of scalp marks or injury. Cranial imaging studies were obtained if clinically indicated. There were 28 infants with scalp trauma, including 17 superficial lacerations, six large caputs, and 12 cephalohematomata; one infant had subgaleal, subdural, and subarachnoid hemorrhages. Logistic regression analysis showed duration of vacuum application to be the best predictor of scalp injury, followed by duration of second stage of labor and paramedian cup placement. Duration of vacuum application ranged from 0.5 to 26 minutes, with a median length of 3 minutes. The proportion of injuries was greater for applications exceeding 10 minutes (6 of 9) than for those 10 minutes or shorter (22 of 121, P < .01). We did not encounter any cases of clinically important scalp injury. Cosmetic scalp trauma occurred in 21% of our newborns delivered by vacuum extraction and was more common after longer vacuum applications, longer second stages, and paramedian cup placement.