Clinical data from two hundred consecutive patients undergoing surgical procedures at the Salt Lake City VA Hospital form the basis of this study. Results of nine commonly ordered preoperative tests (blood count, differential, electrolytes, chemistry panel, urinalysis, prothrombin time, partial thromboplastin time, electrocardiogram, and chest X-ray) were matched with the preoperative history and physical examination and the outcome of surgery in each patient. Each test was examined by the frequency with which it was ordered, the frequency with which it was abnormal, and the frequency with which the abnormal result affected preoperative care. A prevalence of medical illness was found in this population, with 47.5% having a major cardiovascular diagnosis, 35.5% a metabolic or endocrine disease, and 28% a major pulmonary diagnosis. A total of 1271 tests were performed, with 477 (35.5%) showing some abnormality. However, only 76 (5.9%) changed the patient's management before surgery. All but five of these abnormalities were predictable from the clinical evaluation and these five were minor. The overall postoperative complication rate was 9%. No complication was attributed to the omission of a preoperative test. No surgical cases were cancelled during the study period based solely on a preoperative test. We conclude that many preoperative tests can be safely eliminated by ordering only those based on a specific abnormality in the history or physical examination, resulting in more cost-effective surgical care.