The Brazilian equivalent of the Attorney General, as a brief historical analysis shows, plays an essential role in concentrated constitutional review. However, empirical studies of its performance remain scarce. Our goal is to test the AGs track record in Direct Actions of Unconstitutionality (ADIs) at the Brazilian Supreme Court. We use a data set obtained from the Supreme Court in Numbers project’s database in order to run regressions with different models. The hypothesis we tested is that there is no statistically significant relationship between the performance of the AG as plaintiff in concentrated constitutional review and the outcome of the cases. This hypothesis was found to be disproved: ADIs started by the AG, as well as those proposed by representatives of the Executive branch, have, in fact, greater statistical probability of success.