The DNA content of 203 cases of rectal cancer, with at least 15 years follow-up, was analysed by flow cytometry. One hundred and twenty-nine (64%) were DNA aneuploid and corrected survivals were significantly influenced by ploidy distribution (p less than 0.01). The DNA content and details of stage and grade were subjected to multivariate analysis using the Cox regression model. Ploidy was entered into models including Dukes' stage alone, tumour differentiation alone, Dukes' stage and differentiation in combination and a more comprehensive range of discrete stage- and grade-related parameters. All four models demonstrated its independent contribution to survival. However, its contribution was very small (5%, 18%, 5%, 4% respectively). These findings illustrate that results based on new technological developments should not be viewed in isolation, but their values must be assessed in combination with traditionally available data by means of multivariate analysis.