The parole system in Macao was in recent years object of criticisms against its discretionary nature and in particular against the way decision-makers determine who’s coming out earlier and who’s staying longer inside the prison cells. Basically, those criticisms condemned the presumed absence of a pattern in decisions, the opacity of criteria, and the uncertainty in outcomes. This thesis addresses the topic of parole, and focuses on the stage of release decisions. Through a quantitative analysis of prison recommendations for all the 752 cases at first consideration from 2008 through 2010, the study attempts to determine whether there is a pattern emerging from data, and which are the most salient criteria in those recommendations. A theoretical framework inspired on the theory of ‘focal concerns’ was adapted to the specific context of decision-making by prison administrators. Bivariate and multivariate statistical techniques were used to analyze data. Results provided evidence of a clear and very strict pattern in prison recommendations. In consequence, the presumed unpredictability in release decisions was not confirmed. In effect, it was found that a small number of factors, and especially two of them – institutional misconduct and prior convictions – predict in a great extent the outcome of prison recommendations at first consideration. Findings revealed that parole is primarily used as a mechanism of risk management, for controlling: (a) risks associated with future criminal behavior, and (b) risks associated with the breakdown of prisoners’ discipline. In consequence, when making decisions on conditional release, prison managers are mainly concerned with the protection of the community and the maintenance of internal order and security