Developmental reading disability (dyslexia) has traditionally been attributed to impaired linguistic skills. Recent psychophysical data suggest that dyslexia may be related to a visual perceptual deficit. A few visual evoked potential (VEP) studies have addressed this hypothesis, but their results are far from consistent. We submitted 9 dyslexic subjects and 9 age- and sex-matched normal controls to checkerboard pattern reversal VEPs. The main experimental variables were: large (0.5 cycles per degree; cpd) and small (2 cpd) checks and two reversal frequencies (2.1 Hz and 8 Hz); mean luminance and contrast (60 cd/m2 and 50%, respectively) were kept constant in all four conditions. Transient VEP (2.1 Hz) parameters did not differ between controls and dyslexics at 2 cpd. At 0.5 cpd, N70 amplitude was significantly smaller and N70 latency significantly shorter in dyslexics. Amplitudes for the fundamental frequency (8 Hz), as well as for the second and third harmonics of the steady-state VEPs were smaller in dyslexics for both stimulus sizes. A discriminant analysis correctly classified each subject. Our data confirm the hypothesis of a perceptual deficit in dyslexic subjects. The abnormalities are related to spatial and temporal stimulus frequencies: they appear when large stimuli are presented, or when the stimulation frequency is high. These data support the hypothesis of selective magnocellular dysfunction in dyslexia.