The neurofunctional effects of Cognitive training (CT) are poorly understood. Our main objective was to assess fMRI brain activation patterns in children with ADHD who received CT as an add-on treatment to stimulant medication. We included twenty children with ADHD from a clinical trial of stimulant medication and CT (10 in medication + CT and 10 in medication + non-active training). Between-group differences were assessed in performance and in brain activation during 3 fMRI paradigms of working memory (N-back: 0-back, 1-back, 2-back, 3-back), sustained attention (Sustained Attention Task - SAT: 2 s, 5 s and 8 s delays) and inhibitory control (Go/No-Go). We found significant group x time x condition interactions in working memory (WM) and sustained attention on brain activation. In N-back, decreases were observed in the BOLD signal change from baseline to endpoint with increasing WM load in the right insula, right putamen, left thalamus and left pallidum in the CT compared to the non-active group; in SAT - increases in the BOLD signal change from baseline to endpoint with increasing delays were observed in bilateral precuneus, right insula, bilateral associative visual cortex and angular gyrus, right middle temporal, precentral, postcentral, superior frontal and middle frontal gyri in the CT compared to the non-active group. CT in ADHD was associated with changes in activation in task-relevant parietal and striato-limbic regions of sustained attention and working memory. Changes in brain activity may precede behavioral performance modifications in working memory and sustained attention, but not in inhibitory control.