In order to define which patients may not require a routine preoperative chest X-ray, a prospective multicenter study was carried out. It included 3,959 consecutive fifteen, or more, year-old patients, free from any cancer, scheduled for a general or gastrointestinal surgical procedure other than thoracotomy, and had a plain chest X-ray beforehand. This investigation was prescribed before surgery, either by the surgeon or the anaesthetist. Patients were classified according to selected risk factors: age, smoking, emergency surgery, a past history of lung, heart or vascular disease, abnormal clinical findings related to the cardiovascular and respiratory systems, and a previous chest film made less than one year before. There were 2,092 patients in Group I (no risk factors), 916 in group II (one risk factor), 645 in Group III (two risk factors), and 276 in group IV (three risk factors). Three endpoints were selected: a modification of operative schedule or anaesthetic technique, a change in surgical procedure, and the diagnosis of postoperative complications. A rate of 23.2% of preoperative chest films were considered to be abnormal. This rate increased with age and the number of risk factors: 6.2% in Group I and 72.5% in Group IV. Surgical and anaesthetic procedures were modified as a result of the chest X-ray in only 0.5% of patients: 0.1% in Group I, 0.3% in Group II, 1.2% in Group III and 1.4% in Group IV. When pulmonary or cardiac complications did occur after the surgery, the preoperative chest film was of no help for making this diagnosis in more than 50% of cases.(ABSTRACT TRUNCATED AT 250 WORDS)