The lack of clarity about the definition of Computational Thinking (CT) undermines its assessment and the formulation of effective learning strategies for its development. We propose an assessment tool in Spanish, designed by carefully selecting Bebras tasks, to measure four specific CT skills in first-year university students with no prior programming experience. The test was validated with a sample of 980 students from a Latin American university. We found a high, positive correlation between test results and Math test results from the National University Selection System, and a medium, positive correlation between test results and grades from the midterm exam in a Programming course. An acceptable level of internal consistency was found (Cronbach's alpha=0.70). We examined the validity of the test using Classical Test Theory. One question showed poor discrimination potential. Its elimination from the test increased the internal consistency, so we propose to replace it.