Evidence of an association between poor early number skills and a high risk for later difficulties in mathematics manifests a need for both early identification of risk and effective prevention of difficulties. However, only a few theory-based assessment and progress monitoring tools, and intervention methods are avail able for this purpose in Finnish pre-primary education. This thesis aimed to examine to what extent a theory-based computer-assisted intervention method (GraphoGame Math, GGM) can support early number skills development in children (6௅7-year-olds) most at risk for mathematics difficulties (MD; defined as performance below the 10th percentile). The specific aims were threefold: to examine, first, the immediate and delayed condition-specific effects of an intensified GGM intervention; second, the potential effects of sample characteristics on gain scores; and third, the effects of two theory-based intervention sequences on basic addition skills. The thesis is based on three intervention data sets collected during the years 2007–2011. The first data set comprised 17 (out of a total of 236 children), and the second data set 21 intervention participants (of approximately 350 children) identified as having poor early number skills. The third data set comprised 33 intervention participants identified as having a need for extra support for their early number skills (Examination 1), and 14 participants identified as having poor early number skills (Examination 2) (out of a total of 278 children). The results revealed: 1) positive, condition-specific, immediate and delayed intervention effects after intensified GGM practice on dot counting, verbal counting, composing, and basic addition skills; 2) individual differences in response to intervention among the children most at risk for MD, while the benefits for low-achieving children were more general and transferable; and 3) specific effects of number concept training on basic addition skills. The findings suggest that, theory-based computerized methods with game-log data analyses can be successfully used in systematic progress monitoring, remediation of initial gaps, and ongoing identification of difficulties. The data sets were collected as part of the LukiMat-project coordinated by the Niilo Mäki Institute and funded by the Finnish Ministry of Education and Culture since 2007.