Sixteen ADHD children and a control group were asked to reproduce the varying time duration of successively presented visual stimuli. Time estimation was poorer in ADHD children, who showed more impulsive errors. ERPs exhibited similar grand-mean waveforms for both groups during the estimating period, but they were significantly different during the reproducing stage, when an early positive wave over frontal regions characterized the control group, interpreted as memory-guided motor output, followed by a slow negativity probably reflecting an inhibitory motor closure process, both probably involving central executive networks that seem to be improperly activated in ADHD children.