Temperament is one of the factors that explain individual differences in child cognition, and it is usually measured through parental reports. Few studies showed that temperament reports varied depending on whether the reporter is a parent or a teacher, but a quite unexplored question is how each report predicts child cognition. This study analyzes how temperament reports from parents and teachers, predict the performance in a fluid intelligence task in 87 Argentinean children aged 4-5 years from medium-to-low socioeconomic status. Children were administered the Matrix subtest of KABC-II, and their temperament was evaluated with the CBQ Very Short Form, administered to parents and also to teachers. We will run one multilevel mixed-effects linear regression for CBQ dimensions, including teacher report of child temperament as the first level of nesting, and parental report of child temperament as the second level; the Matrix subtest’s total score will be the dependent variable.