In a clinical judgement analysis study, 27 general practitioners, 22 cardiologists, and 21 medical students assessed 40 case vignettes with regard to the probability of heart failure, in order to study the weights of different kinds of information (cues) measured by the regression coefficients in a multiple regression model. The vignettes were based on actual patients. We found that diagnostic accomplishment and diagnostic strategies were surprisingly similar on the group level, but very different on the individual level. The most important cues for the participants were cardiac enlargement and pulmonary stasis. Strategies in which cardiac enlargement was the predominating cue led to a higher diagnostic accomplishment; a third of the participants used such strategies. The cues given in the vignettes could have been utilized more efficiently; cardiac enlargement seems to be more important and "classical" symptoms less important for predicting heart failure than the participants realize.