ICT infrastructure investments in educational institutions have been one of the key priorities of education policy during the last decade. Despite the attention, research on the effectiveness and efficiency of ICT is inconclusive. This is mainly due to small-scale research with weak identification strategies which lack a proper control group. Using the 2011 ‘Trends in International Mathematics and Science Study’ (TIMSS) data, we define by a Mahalanobis matching a control group with similar student, teacher, school and regional characteristics. The results indicate that accounting or not accounting for these characteristics, may considerably alter the estimated impact of ICT. This suggests that a correction for characteristics related to the student population, teaching staff, administrative personnel and school management is warranted in the evaluation of the impact of ICT.