AIM: To investigate the possibility that the incidence of apoptotic bodies in the cryptal epithelium might help to identify colonic lesions due to drugs, especially non-steroidal anti-inflammatory drugs (NSAIDs). METHODS: The apoptotic count (AC) the number of apoptotic bodies per 100 crypts was calculated in a series of colorectal biopsy specimens, stained with haematoxylin and eosin from patients with (a) known or suspected drug induced colitis and (b) inflammatory bowel disease before or after treatment with salazopyrine or corticosteroids. These specimens were compared with normal biopsy specimens from a control group of comparable age and sex distribution. RESULTS: Under normal conditions apoptotic bodies were seldom seen at all and the mean apoptotic count was less than 1.0. In untreated inflammatory bowel disease the mean apoptotic count was marginally increased (2.4), but when there was a partial response to drug treatment the apoptotic count rose to 13.1 (p 0.003). In colonic lesions directly attributable to drugs the apoptotic count was always increased, reaching its highest level (106) with 5-fluorouracil. In colitis related to NSAIDs apoptoses were associated with inflammation, most notably an increase in lymphocytes in both lamina propria and epithelium. CONCLUSION: The presence of crypt apoptoses in substantial numbers (with an apoptotic count in excess of 5) should always raise the possibility of drug effect. The mechanisms involved are not clear but with NSAIDs the changes might well be immunologically mediated.