The present work aims to compare two different risk aversion measures, VaR and CVaR, obtained by the models GARCH and GAS, for three distinct financial return series of the Brazilian stock market. For each series, it was estimated the GARCH model with normal gaussian errors, the GARCH model with t- Student errors and the GAS model with t-Student errors. Each model was evaluated through its standardized residuals analysis. The Value-at-Risk model for each series was validated by the coverage tests to be presented in this work. It was verified that the model with the most adequate coverage was the GARCH model with normal Gaussian errors. The implementation was made using the softwares Matlab and EViews.