Using clinical factors, the probabilities of survival with and without major handicap were separately calculated by multiple logistic regression for 988 children who weighed 2000 g or less at birth and who were born in the period 1976-1980. For survivors weighing 501-1250 g in whom incidence of serious handicap was 21%, neonatal fits and the need for mechanical ventilation carried significantly increased risks of later serious handicaps. In contrast, for survivors weighing 1251-2000 g in whom the incidence of serious handicap was 6%, significant factors were spontaneous, uncomplicated preterm delivery, recurrent apnoea, and abnormal neonatal neurological findings. The probability of dying was assessed in two ways--firstly, using seven clinical factors available on admission to the neonatal unit, and secondly, using 10 perinatal and neonatal factors. Handicapped survivors had a perinatal risk between that of those babies who survived and that of those who died. Of the handicapped survivors, those with spastic diplegia and hemiplegia had been extremely low risk babies (medians 1.4% and 1.6%, respectively), whereas those with other impairments had much higher risks (range 17.5-38.1%). We postulate that certain impairments arise independently of clinical events, although most occur in children who had complicated perinatal courses.