Objective.: To investigate effects of methylphenidate (MPH) on numerical performance in children with Attention- Deficit/Hyperactivity Disorder (ADHD) with and without concurrent math difficulties. Method.: Data were analyzed from three groups of children with ADHD, who varied in arithmetic abilities. Groups were matched for IQ and reading abilities and classification was based on ICD-10 criteria, using scores on a standardized arithmetic achievement test. Thus, we identified one group with severe difficulties in arithmetic (ADHD+Developmental Dyscalculia; DD), second group with more general and less severe difficulties in arithmetic (ADHD+Mathematical Disabilities; MD), and a third group with good arithmetic abilities (ADHD). All children completed a 10-minute arithmetic task involving subtraction problems, during an acute, randomized, placebo-controlled cross-over trial with three dose levels of methylphenidate (10mg, 15mg, 20mg). Results.: ( 1 ) Both ADHD+MD and ADHD+DD were impaired in using strategies that implicate working memory (i.e., borrowing). However, only the ADHD+DD were impaired in using implicit knowledge of quantities (i.e., doing simple subtractions). ( 2 ) MPH improved all children’s performance of arithmetic procedures (borrowing) that involves working memory, but had no effect on basic numerical skills that involves understanding of quantities. Conclusions.: We show clear dissociation of MPH functions: it improves working memory functions but does not improve specific cognitive functions such as quantity manipulation. Moreover, MPH shows decreased efficacy for arithmetic performance in ADHD+DD, highlighting the need for additional intervention in this subgroup.