There is extensive literature on the effects of levodopa treatment on the visual evoked potentials (VEP) in laboratory animals and in depressed patients. The effects of levodopa on the VEP of parkinsonian patients were overlooked to a certain extent. In this work we searched for levodopa effects on the VEP of 42 parkinsonian patients. The VEP were data-reduced and analyzed by several techniques: (a) 256 data values (256 msec) were reduced to 70 variables by averaging; (b) time-domain parametric extraction: latencies to peak-and-trough points, amplitudes, etc., resulted in 14 variables; (c) frequency transformation into power spectral density bands resulted in 14 variables. Each one of the above variables was univariate, paired t-tested for levodopa effects. The overall effects of levodopa on the power variables and the time-domain parameters were evaluated by Hotelling paired T2. The 70 time variables were further reduced by the tolerance function of the discriminant procedure and analyzed both by direct and by stepwise discriminant analyses and by SAS-MANOVA in a pairwise design. Only few sporadic univariate t values reached significance levels. No overall multivariate significant effects of levodopa were found. In view of known dopaminergic involvement in parts of the visual system it is postulated that levodopa might have antagonizing effects at different levels of the visual pathways. This hypothesis and better understanding of levodopa effects on the visual system should probably be achieved by recording VEP from the visual subsystems.