The DNA distribution pattern was determined retrospectively in 25 rectal carcinomas and the possible correlation to clinical outcome evaluated. The DNA content in individual cells was measured according to a cytophotometric method based on light transmission measurement of Feulgen-stained nuclei. Tumor cells with DNA content exceeding an upper limit, i.e., the 90 percentile of the control cells, were considered to be nondiploid (aneuploid). Virtually all long-term survivors had less than 50 per cent of the tumor cells exceeding the upper diploid level, whereas those developing only a local recurrence had 50 to 70 per cent. Patients with disseminated disease and short survival time had all of their tumor cells exceeding the upper diploid level. There was a highly significant correlation between Dukes' stage and aneuploidy and probably a significant correlation between histologic grading and aneuploidy. The clinical significance of these results lies in the fact that DNA can be measured in biopsy specimens. It might thus be possible to "tailor" the operation according to the future clinical course to be expected. It could be hypothetically argued that patients with a DNA profile heralding disseminated disease and short life expectancy should have surgery that preserves quality of life, whereas those tending to develop a local recurrence should have more aggressive surgery. It may also be possible to define groups of patients thought to prosper from a more intense postoperative surveillance. The scientific basis for these suggestions is still lacking, and further studies on a prospective basis are currently in progress.