Carcinoma of the anal canal is an uncommon malignancy, accounting for only 1-3% of colorectal cancers (Goligher, 1984). Flow cytometric DNA analysis has attracted interest as a possible prognostic tool in patients with colorectal cancer (Wolley et al., 1982; Armitage et al., 1985;Kokal et al., 1986;Quirke et al., 1987; Scott et al., 1987a, b tional 28 patients had been fixed continuously in formalin for a penrod of several years. As the tumour tissue of these 28 patients would not stain satisfactonrly with propidium iodide (Schutte et al., 1985), they also had to be excluded from the study.A DNA histogram was obtained from the anal cancers of 135 patients. DNA histograms with a single GO GI peak were classified as DNA diploid. A DNA diploid ploidy pattern was given by 82 anal canal cancers (cv of GO/GI peak 7.95%).DNA histograms that contained an additional GO/GI peak with a DNA index (DI) value of between 1.10 and 1.99 were classified as DNA aneuploid (Hiddemann et al., 1984); this latter pattern was seen in 15 anal cancers. Finally, DNA ploidy patterns with an especially large G2 peak (mean DI 2.10; s.e.m. 0.05) containing more than 10% of the measured cell nuclei (Tribukait et al., 1982). in the absence of aggregation (no 6c peak), were classified as DNA tetraploid; this category of DNA tetraploid included 20 anal cancers. The DNA histograms of the remaining 18 anal carcinomas were of such poor quality that they could not be classified.Statistical analyses were carred out using the SAS procedures (SAS Program Institute, 1986). The program FREQ was used to test associations of clinicopathological features of the anal canal cancers with their DNA pattern. All P values were determined using the Pearson X2 statistic. The survival curves were generated using the Kaplan-Meier method (Kaplan & Meier, 1958), and univariate survival comparisons were made using the log-rank statistic (Mantel & Haenszel, 1959 (Table I). Similarly, little correlation was seen between tumour ploidy and patient age, although younger patients (less than 45 years) had proportionately more DNA non-diploid patterns (Table I). Large anal cancers (greater than 5cm diameter) gave more DNA non-diploid patterns than smaller anal cancers (Table I).