A prospective study was undertaken to determine the safety of the Silastic vacuum extractor. Between November 1982 and July 1983, a cohort of 84 successful vacuum extractor deliveries was examined, using the next sequential forceps delivery and spontaneous vaginal delivery as controls. In addition to routine neonatal morbidity measures, Scanlon early neonatal neurobehavioral scale and a modified Sarnat encephalopathy staging examination were used to critically assess neurologic functioning; a cranial ultrasound scan was performed to look for intracerebral hemorrhage, and an indirect ophthalmologic examination was done to assess the incidence of retinal hemorrhage. The study yielded no significant increase in maternal vaginal trauma for vacuum extractor versus spontaneous vaginal delivery, but there was a significantly greater incidence for forceps delivery (60%) versus vacuum extractor (25%) and more associated blood loss for forceps delivery (P less than .01). There was no significant increase in neonatal morbidity for vacuum extractor compared with forceps delivery nor in serious morbidity compared with spontaneous vaginal delivery. Specifically, for vacuum extractor versus forceps delivery there was no difference in one- and five-minute Apgar scores, extent of resuscitation, cosmetic injury, jaundice, mean neonatal intensive care unit stay, or incidence of retinal hemorrhage. Notably, there was no mortality related to delivery method, but there were two unrelated deaths. There were no cases of intraventricular or subgaleal hemorrhage on clinical or ultrasound examination, but one stillborn infant, who succumbed to a generalized coagulation defect, had a subarachnoid hemorrhage. Finally, there was no significant difference in Sarnat encephalopathy staging or Scanlon neurobehavioral assessment between spontaneous vaginal, forceps, and vacuum extractor deliveries.(ABSTRACT TRUNCATED AT 250 WORDS)