We performed a decision analysis to evaluate cost per cancer detected, cost per neoplasm detected, and cost per treatable lesion of two common diagnostic strategies, barium enema-proctoscopy or colonoscopy as the first diagnostic test, for patients with fecal occult blood loss. The prevalence of polyps, cancer, and angiodysplasia, and the colonoscopy success rate were obtained from consecutive colonoscopy records. Costs were estimated from hospital charges; sensitivity and specificity of barium enema and colonoscopy were obtained from the literature. For treatable lesions (cancer, polyps, and angiodysplasia), the colonoscopy first strategy had a higher sensitivity (80% vs. 57%) and a higher specificity (95% vs. 80%) than the barium enema first strategy. Cost effectiveness measures were similar for the two strategies. Colonoscopy as the first diagnostic test had a lower cost per treatable lesion ($2,319 vs. $2,895) and a lower cost per neoplasm detected ($2,694 vs. $2,896), whereas the barium enema first strategy had a lower cost per cancer detected ($10,050 vs. $10,297). The lower cost per treatable lesion of the colonoscopy first strategy was not affected by changes in the prevalence of lesions, test characteristics, costs of tests, or colonoscopy success rate over clinically relevant ranges. The higher cost of colonoscopy was offset by its greater sensitivity and its capacity for biopsy and therapy. Therefore, since the cost per treatable lesion is lower and the sensitivity, specificity, and predictive value is superior, colonoscopy is recommended as the preferred initial test in evaluating a patient with fecal occult blood loss.