Adverse events in organizations are more than a serious concern. Over the last few years the awareness of this problem has raised and different organizational solutions have been tried. We focus on the problem of managing operational and clinical risks, in terms of events that influence the success of service delivery. This paper is aimed at proposing risk management as the basic methodological approach to deal with adverse events and risks. We propose Bayesian networks (BNs) to assess risk profiles given a context of application and benchmarks by Bayesian decisional theory to evaluate the profiles, i.e. defining the acceptability of them. The method is described both at a theoretical and an empirical level, thanks to its application to health care (haemodialysis department) and banking field. The occurrences of these top events are modeled by Bayesian networks which gather posterior risk profiles for each patient or banking business line. The comparison of them with a reference risk profile is input for decision making. BNs augmented with decisional nodes and scenario analysis complete the risk management process. The ultimate goal is to improve risk profile and, consequently, service supply quality in the organization.