Diagnostic strategies applied in 144 consecutive patients with jaundice were investigated. Diagnostic hypotheses formed on the basis of initial data--history, physical findings, and results of routine laboratory tests--enabled the clinician to adopt a single-target strategy in 75 patients and a multi-target strategy in 61 patients. Four patients died very early in the course of events and another four were excluded from analysis because they were judged not to benefit from further investigations. The single-target strategy, in which the clinician explored one diagnostic hypothesis only, often led to a short diagnostic process (mean, 11 days), and in 89% of the patients the clinician's hypothesis was correct. In the multi-target strategy several diagnostic hypotheses were evaluated. In these patients the correct diagnosis was included in the first set of hypotheses in 84%, and the mean duration of the diagnostic process was 25 days. Patients with cholestatic jaundice presented the most difficult diagnostic problem, and most of these were investigated by a multi-target strategy. The possibility of an extrahepatic obstruction often forced the clinician to use invasive procedures to rule out this diagnosis. Considerable time was spent in observing the clinical course and waiting for investigations to be carried out. To reduce the investigative cost, controlled studies of diagnostic value and optimal order of investigations are warranted.