Computational thinking (CT) has been recognized as an essential part of every child's education (Wing, 2006) and Bebras contest items have been frequently used to measure CT. Think alouds has emerged as a prominent method for identifying thought processes and correcting problems with assessments. As little research has examined validity evidence of Bebras items, this study conducted both quantitative and qualitative think alouds validation of selected Bebras items. The results showed that the selected Bebras items significantly correlated with another CT test. Students' think alouds revealed that some Bebras items measured construct-irrelevant skills and the proposed CT components of each item are not consistent with student thinking. This study proposes an effective validation process for CT assessments.