The purpose of this study was to identify clinical characteristics that could predict the diagnosis in ambulatory patients with abdominal pain. We studied 552 unselected ambulatory male patients whose average age was 47 years and whose median duration of pain was 3 weeks. Potentially serious disease occurred in 21% of the patients. Single abnormal findings had a low predictive value for serious disease. However, by using combinations of clinical findings, we could construct and test a decision rule to identify a group of patients who had a low prevalence of serious disease. This "low risk" group contained 36% of all patients with abdominal pain. Laboratory tests were almost always normal in these patients. Our findings suggest a diagnostic strategy for evaluating abdominal pain: When the initial examination shows that there is little chance of serious disease, laboratory tests should be deferred or omitted altogether. In patients who have a very low likelihood of potentially serious disease, it may be useful to regard "nonspecific abdominal pain" as a positive diagnosis, rather than a diagnosis of exclusion.