Several studies indicate long-term cognitive impairment of MDMA ( ecstasy) users. In the present study we attempted to establish whether electrophysiological correlates of low-level cognitive processes present a long-term alteration, dependent on the level of use of ecstasy. We addressed this issue by investigating amplitude and latency of VEPs related to a very simple discrimination task involving sustained attention ( arousal). Eight heavy-MDMA users, eight moderate-MDMA users and 18 drug-free control subjects were asked to discriminate whether the digit at the centre of the screen was 1 or 2. None of the subjects ( except one) had used MDMA in the 6 months previous testing. We measured psychophysical performance and EEG, recorded in Oz and Fz during task execution. The heavy-MDMA users made significantly more errors than the other two groups ( p<.05). Moreover, they presented reduced amplitude but not latency of VEPs in both Oz and Fz. The effect in Oz is present in P200 ( for heavy users only, p<.05) and in P300 components ( for both MDMA groups; heavy users: p<.001, moderate users: p<.0.5). In Fz, the amplitude effect is present in N250 ( for heavy users only, p<.05) and in P300 components ( for both MDMA groups; heavy users: p<.05, moderate users: p<.05). The three groups do not differ in early components, reflecting low-level processing. These results provide evidence of long-term electrophysiological abnormality displayed by ecstasy users and agree with the suggestion that even typical recreational doses of ecstasy are sufficient to cause long-term altered cortical activity in humans.