The development of a group of 100 preterm infants, mean birth weight 1,877 grams, from a broad range of social class and ethnic backgrounds was followed from birth to age 5. Assessments in infancy were directed at medical problems and early perceptual, motor, social, and cognitive development. The child's performance on the Stanford-Binet test at age 5 could not be predicted from early hazardous events in the obstetrical or neonatal period. The results indicate that developmental outcome at age 5 could be predicted moderately from a single measure, infant visual attention, administered as early as term date. Prediction was improved by using a combination of assessments given during the first 9 months. Furthermore, prediction was significantly better for girls than for boys. Although moderate stability in performance was found for the group as a whole, prediction of an individual's performance resulted in a substantial number of children being misclassified. Social factors were more important than any other set of factors in relating to the child's mental performance at age 5.