In a study of simulated consultations with patients suffering from dyspepsia, the amount and nature of information required by 20 general practitioners to make management decisions about these simulated patients were analysed. The number of items of information required to complete all eight cases ranged from 23 to 128 (median 66). Doctors who consulted more rapidly in actual surgeries required less information in the simulations (P less than 0.01). Two-way analysis of variance showed that both cases and doctors independently influenced the number of items of information collected, but that doctors were a more powerful influence (P less than 0.05). The nature of the information collected by doctors varied widely between individuals. There was no correlation between the professional experience of the doctors studied and the amount or nature of data collected; this is at odds with previous studies and requires an alternative explanation.