Vigabatrin (VGB), a treatment for the childhood epilepsy, infantile spasms (IS), is implicated in visual field constriction. Electroretinograms (ERGs) are used as a substitute for visual field testing in infants. We use the VGB-associated ERG reduction (VAER), defined as reduction in age-corrected light adapted 30 Hz flicker amplitude from a pre-treatment measurement in the absence of other retinal defects, as an indicator of retinal toxicity resulting from VGB use. The d-wave ERG response is predominantly the result of OFF-bipolar cell depolarization response to light offset. The purpose of this study is to evaluate the ERG d-wave response as a marker for VAER toxicity in an infant population. One hundred children with IS treated with VGB (median age at baseline: 7.6 months; range 1.7-38.4) were tested for the cone-OFF response elicited to a 250 cd s m(2) flash with 200 ms duration (long flash ERG). Diagnosis of VAER requires baseline testing of the flicker ERG and at least one follow up ERG; Fifty-one patients fulfilled this criteria. Fifty-eight children received the long flash ERG at baseline. Thirteen retinally normal controls with a median age of 32 months (5.7-65) were also tested. Amplitude and implicit time of the d-wave response were measured manually. Longer duration of treatment was associated with reduced d-wave amplitude (ANOVA p < 0.05) in patients taking VGB. Nine patients demonstrated VAER during the course of the study. D-wave amplitude was reduced in the IS group with VAER compared to those without VAER (p < 0.05). Vigabatrin associated retinal defects may be reflected in reduction of the cone d-wave amplitude.