Health Administrative Databases are among the most promising data in healthcare because of the capability to generate population-based knowledge. This study shows that Health Administrative Databases produce significant knowledge for evaluating hospitals’ performances and supporting policy-makers to set improvement strategies. The empirical setting is the Lombardy Region (Italy) and hospitals treating patients affected by Heart Failure(HF). Thirty-day mortality and readmission can be used to identify high-performing hospitals in the case of HF patients. High- and low-performing hospitals were identified as outliers through a funnel plot comparing expected and observed cases of death and re-admission. Expected cases were estimated using a 2-level logistic model. Funnel plot results on mortality showed that 48 hospitals were located outside the 95th percentile. Twenty-nine of these showed an observed mortality higher than the expected. As for re-admission, the number of hospitals “out of control” was equal to 8, equally distributed between the upper and the lower part of the plot. Health Administrative Databases can provide policy-makers and hospital managers with population-based knowledge that complement the traditional sources such as clinical trials and medical registers. While administrative data have been used mainly for epidemiological analyses, this study leverage on them to evaluate hospital performances, in doing so it paves the way for further research. The application of mixed models and funnel plot to detect hospital outliers based on administrative data is unusual and promising. Although our model has been developed for HF patients, we argue that it can be extended easily to other pathologies. Health Administrative Databases can generate robust knowledge about hospital performances. Our results about hospital treating HF patients show that private hospitals are more in-control than public ones. Public hospitals offer both the best and the worst performances. Moreover, hospitals that over-perform in term of re-admission are likely to over-perform also in term of mortality: this paves the way to strategies that could improve both performances at the same time. Finally, being based on a very large basis of evidence, policy-makers will be able to win resistance to change.