OBJECTIVE To review the incidence and severity of retinopathy of prematurity (ROP) in infants with birthweights 1000-1249 g and 1250-1499 g, to establish whether the upper weight limit for routine ophthalmological examination might safely be lowered. METHODS Prospective cohort study of infants born between 1 January 1977 and 31 December 1992 cared for in the neonatal nurseries at the Royal Women's Hospital, Melbourne. Data were retrieved on 1373 infants who survived their initial hospitalization. They comprised 657 with birthweights 1000-1249 g (group 1) and 716 with birthweights 1250-1499 g (group 2). There were 76 outborn infants in group 1 and 97 in group 2; the remaining infants were all born at the Royal Women's Hospital. Ocular examinations commenced at 2 weeks of age, when possible, and at 2-weekly intervals after that. RESULTS In group 1, ROP was detected in 14.6% (96/657) and severe ROP (bilateral stage 3-5) in 5.0% (33/657). Five (0.8%) children required surgical intervention (reaching threshold disease); following surgery, one was legally blind, one had severely impaired vision, and the other three had near-normal vision. Another child was blind; he was born at 28 weeks gestational age with a birthweight of 1170 g, and was transferred to a Level II hospital at 9 weeks chronological age with no detectable retinopathy. He returned 1 year later totally blind with detached retinae (grade 5 ROP). The prevalence of bilateral blindness in this group was 0.3% (2/657). In group 2, ROP was detected in 6.4% (46/716) and severe ROP in 0.8% (6/716). No children required surgery; three were found to be myopic at follow-up but the corrected visual acuity was normal. No children in group 2 were blind. No significant difference was found between the rates of ROP in inborn and outborn infants. CONCLUSIONS In neonatal units with similar rates of ROP and visual outcome, routine ophthalmological examination in the neonatal nursery of infants weighing more than 1249 g at birth is probably unnecessary.