PURPOSE: The visual evoked potentials (VEPs) offer a window both on visual system function and on cerebral or neurologic function. The study was performed to evaluate the maturation of VEP in preterm infants and to present a guide to detect high risk neonates who can develop neurodevelopmental abnormalities. METHODS: The VEPs were recorded in response to binocular stimulus with light-emitting diode goggles. VEP recordings were analyzed in 37 healthy neonates who were followed longer than 6 months and neurodevelopmentally normal. RESULTS: 1) The peak latencies of N1 were 158.4+/-42.4 (msec), 159.4+/-40.6, 80.2+/-20.6, 52.4+/-12.0 from 34 to 42 weeks, respectively. 2) The peak latencies of P1 were 227.7+/-34.4 (msec), 187.0+/-35.4, 124.4+/-30.2, 148.3+/- 44.9 from 34 to 42 weeks, respectively. 3) The peak latencies of N2 were 315.4+/-49.9 (msec), 228.6+/-44.2, 205.5+/-69.3, 194.4+/- 89.9 from 34 to 42 weeks, respectively. 4) The peak latencies of P2 were 403.1+/-87.0 (msec), 275.4+/-61.5, 232.5+/-58.7, 255.1+/- 125.1 from 34 to 42 weeks, respectively. 5) The mean amplitude of N1-P1, P1-N2, N2-P2 were 5.3+/-6.8 (microvolt), 5.3+/-5.5, 5.5+/- 4.1 respectively. The peak latencies were inversely related to post-conceptional age (P<0.05). CONCLUSIONS: The peak latencies of each wave were decreased according to the post-conceptional age and these data reflect maturational changes in brain function.