The purpose of this cohort study was to determine the incidence of and risk factors for major neurodevelopmental impairments among survivors of extreme prematurity. The study cohort comprised 100 infants born between 24 and 28 weeks of gestational age at one tertiary center from 1983 to 1984. Twenty-five infants (25%) died; 75 (75%) survived until follow-up (mean, 60 months). Standardized neurodevelopmental and psychometric assessments were performed in blind fashion on 68 of the 75 surviving children (91% follow-up). Informal assessments (parent, teacher, and physician reports) were obtained instead for seven (9%) children who had relocated outside of the area. Overall, 19 children (25%) had one or more major impairments: mental retardation, 9; cerebral palsy, 4; multiple impairments, 5; and blindness, 1. Despite a high prevalence of impairments, 95% of children (n = 71) were functionally independent [corrected]. Special educational resources were definitely necessary for seven (9%) and possibly needed for 36 (48%) additional children. Univariate analyses revealed four significant risk factors for cerebral palsy: hydrocephalus (relative risk = 12.2), grades III and IV intraventricular hemorrhage (relative risk = 5.8), 5-minute Apgar score lower than 7 (relative risk = 5.7), and bronchopulmonary dysplasia (relative risk = 5.5). Hydrocephalus was the only significant risk factor observed for mental retardation (relative risk = 5.4). Risk factors predicting a need for special education resources included sepsis (relative risk = 24.9), low socioeconomic status (relative risk = 16.3), and nonwhite race (relative risk = 3.0). Thus our data suggest that biomedical factors appear to confer the greatest risk of major impairments; sociodemographic factors appear to have a significant impact on educational risk in extremely premature infants who do not die. Continued follow-up with biomedical and developmental-social interventions appears warranted to decrease the risk of educational underachievement in this population.