Developmental neuropsychology and functional neuroimaging evidence indicates that simple and complex mental calculation is subserved by a fronto-parietal network. However, the effective connectivity (connection direction and strength) among regions within the fronto-parietal network is still unexplored. Combining event-related fMRI and multivariate Granger Causality Mapping (GCM), we administered a multiplication verification task to healthy participants asking them to solve single and double-digit multiplications. The goals of our study were first, to identify the effective connectivity of the multiplication network, and second, to compare the effective connectivity patterns between a low and a high arithmetical competence (AC) group. The manipulation of multiplication difficulty revealed a fronto-parietal network encompassing bilateral intraparietal sulcus (IPS), left pre-supplementary motor area (PreSMA), left precentral gyrus (PreCG), and right dorsolateral prefrontal cortex (DLPFC). The network was driven by an intraparietal IPS-IPS circuit hosting a representation of numerical quantity intertwined with a fronto-parietal DLPFC-IPS circuit engaged in temporary storage and updating of arithmetic operations. Both circuits received additional inputs from the PreCG and PreSMA playing more of a supportive role in mental calculation. The high AC group compared to the low AC group displayed a greater activation in the right IPS and based its calculation more on a feedback driven intraparietal IPS-IPS circuit, whereas the low competence group more on a feedback driven fronto-parietal DLPFC-IPS circuit. This study provides first evidence that multivariate GCM is a sensitive approach to investigate effective connectivity of mental processes involved in mental calculation and to compare group level performances for different populations. Hum Brain Mapp, 2010. © 2010 Wiley-Liss, Inc.