Abstract We report five studies which compared two theories linking surprise to causal attribution. According to the attributional model, surprise is frequently caused by luck attributions, whereas according to the expectancy-disconfirmation model, surprise is caused by expectancy disconfirmation and stimulates causal thinking. Studies 1 to 3 focused on the question of whether surprise is caused by luck attributions or by unexpectedness. In Studies 1 and 2, subjects had to recall success or failure experiences characterised by a particular attribution (Study 1) or by low versus high surprisingness (Study 2), whereas in Study 3, unexpectedness and luck versus skill attributions were independently manipulated within a realistic setting. The main dependent variables were unexpectedness (Studies 1 and 2), degree of surprise (Studies 1 and 3), and causal attributions (Study 2). The results strongly suggest that surprise is caused by expectancy disconfirmation, whereas luck attributions are neither sufficient nor necessary for surprise. Studies 4 and 5 addressed the question of whether surprise stimulates attributional thinking, again using a remembered-incidents technique. The findings of the previous studies were replicated, and it was confirmed that surprising outcomes elicit more attributional search than unsurprising ones. Additional results from Study 5 suggest that causal thinking is also stimulated by outcomes that are both negative and important.