The accumulating evidence suggests that aspirin or other NSAIDs may prevent or inhibit the development of colon and perhaps other digestive tract cancers. Although the clinical, experimental, and epidemiologic evidence is promising, the hypothesis remains unproven except in the models of chemically induced colon cancer in rodents and adenomatous polyps in patients with FAP. Clinicians should await the results of randomized trials before using NSAIDs for cancer prevention or treatment. Recommendations are as follows: 1. Experimental studies should define the mechanism or mechanisms by which NSAIDs inhibit tumorigenesis in the rodent model. 2. Experimental and clinical studies should define the optimal drug, dosage, and treatment regimen. The new, selective COX-2 inhibitors should be studied for efficacy and toxicity. 3. Epidemiologic studies should continue to explore the issues of dosage, duration, drug, and toxicity. Because full-scale, randomized trials are feasible only for studying intermediate end points such as polyp recurrence or proliferative indices in high-risk populations, epidemiologic studies have an ongoing role. 4. Carefully designed randomized, clinical trials, now underway, are needed to test the efficacy of NSAIDs in inhibiting colorectal polyps or cancer in humans. 5. Better criteria are needed as to who should take aspirin and who should not.