Twenty general practitioners co-operated in a study of the decision-making process in general practice respiratory illness. Ten simulated standard-patient consultations were used, each being included twice, once to make a decision on diagnosis and once to make a decision on management.Comparison of simulated behaviour with previous studies of observed and self-recorded behaviour showed encouraging agreement. Examination of 200 pairs of management and treatment decisions showed that fewer pieces of information were required for a management than a diagnostic decision on 111 occasions; in 43 pairs the quantity required was the same for each decision. The exact information required to make a decision on diagnosis was included in the information required to decide management on only 30 (15%) of the 200 possible occasions.When disagreement was found over management of a simulated case a decision to prescribe an antibiotic required less information than a decision not to prescribe an antibiotic.