One hundred and three colorectal carcinoma specimens were examined to determine the value of the xylene and alcohol fat clearance technique in detecting lymph node metastases. The mesocolon or mesorectum was dissected initially by the traditional method and all the lymph nodes identified were examined histologically. After fat clearance the specimen was dissected again and further lymph nodes were examined. Forty-one specimens were obtained from the rectum and 62 from the colon. Traditional dissection produced a mean of 6·2 lymph nodes per specimen, but following fat clearance a further mean of 12·4 nodes per specimen were found. The total number of lymph nodes recovered varied from two to 69 with a mean of 18·5 per specimen. Traditional dissection showed 45 specimens (43·7 per cent) to have lymph node metastases but after fat clearance a further five specimens (4·8 per cent) were found to be lymph node positive. Therefore, of the 58 specimens graded initially as Dukes' B, five (8·6 per cent) were shown after fat clearance to be Dukes' C tumours. In the Dukes' C cases the mean (s.d.) number of involved lymph nodes per specimen was 2·7 (2·1) by traditional dissection and 4·2 (3·9) after fat clearance. Forty-seven (94·0 per cent) of the Dukes' C tumours were correctly identified after examination of specimens containing up to 13 lymph nodes. Fat clearance of the mesocolon or mesorectum should be used when traditional dissection has failed to identify at least 13 nodes and the tumour has been classified as Dukes' B.