BACKGROUND. Outcome bias is a psychological tendency to use outcome information in order to evaluate the quality of a previous decision, where decisions with bad outcomes are undervalued and decisions with good outcomes are overvalued. Previous studies have shown that the bias can be present in some evaluations of diagnostic and/or treatment medical decisions (e.g., Baron and Hershey, 1988; Gruppen, Margolin, Wisdom, Grum, 1994). Sacchi and Cherubini (2004) recently showed that outcome bias can affect doctors evaluating their own decisions, and argued that, in some circumstances, this tendency might have consequences on spontaneous learning: bad diagnostic decisions resulting, by chance, in good outcomes might be improperly repeated in the future, while good diagnostic decisions with bad outcomes might be improperly abandoned. METHODS. A total of 49 practising Italian doctors and nurses were contacted at three times. On the first meeting, the nurses were given a clinical case that required to send out an ambulance, and were asked to decide whether to send a doctor with the ambulance or not, and the doctors were given a clinical case and were asked to choose between two possible diagnoses. Participants had to make their choice and to report their confidence toward its quality. At the second meeting (2 or 3 weeks later), participants were presented with a randomly selected positive or negative outcome of their previous choice. Participants were asked to report their current confidence toward the quality of their previous choice. At the third meeting (2 or 3 weeks later) participants were given a clinical case superficially different but structurally and clinically equal to the previous one, and were asked to choose between two alternatives and to report their confidence toward their current choice. RESULTS. At the second meeting, participants in the positive outcome group showed an increase in confidence, whereas participants in the negative outcome group showed a decrease in confidence, replicating the typical outcome effect. More important, reliably more participants made a different choice in the second clinical case in the negative outcome group than in the positive outcome group. CONCLUSIONS. The results show that outcome bias can affect spontaneous learning. After a negative outcome clinicians changed their decisions on a structurally equal case more than after a positive outcome.