The reasons why Brazilian civil engineering education has not achieved competitive performance worldwide remain unclear. Thus, this study aims to identify what factors influence student performance in civil engineering undergraduate programs in public institutions in Brazil. Multiple linear regression methods were used to analyse the interactions between the student performance (dependent variable) and the professors' and universities' attributes (predictors). The analysis shows that the professor's qualification and didactics are significant attributes, as are the time of the program's existence, the academic organisation (if it is annual or semester courses), and the institutions' administrative jurisdiction. The discussion presented possible explanations for each attribute's contribution. The study adds to the theory of predicting student performance in a developing country scenario, in this case, Latin America, where engineering education has less investments in academic infrastructure and human capital.