Summary To examine the hypothesis that colorectal carcinomas with and without TP53 mutations may be characterised by aetiological heterogeneity, we analysed a group of 107 patients with primary Dukes' C colorectal cancer seen at the Memorial Sloan-Kettering Cancer Center (MSKCC) from 1986 to 1990. We assessed p53 overexpression using the monoclonal antibody PAb 1801, and identified 42 (39%) patients displaying p53-positive phenotype, defined as >25% of positive cells. Patients with two or more first-degree relatives with cancer had an odds ratio (OR) of 2.9 (95% CI 1.0-8.3) for p53 overexpression in comparison with those without a family history of cancer (trend test, P = 0.11). A possible association between body weight and p53 overexpression was observed. The ORs were 1.9 for the second quartile, 1.9 for the third quartile and 3.4 for the highest quartile in comparison with the lowest quartile (trend test, P = 0.06). No association between occupational physical activity, smoking, drinking, parity and p53 overexpression was identified. The results suggest that p53 overexpression may be related to genetic predisposition to colorectal cancer, and p53-positive and p53-negative colorectal cancers may be controlled by different aetiological pathways.Keywords: body weight; colorectal neoplasms; family history; p53/protein; risk factors Colorectal cancer (CRC) is the second most common malignancy of both sexes in developed countries . A total of 394 000 deaths are estimated to occur annually for colorectal cancer, making it the third most important cause of cancer mortality in the world . It is also a major public health problem in the United States, with an estimated 149000 new cases diagnosed in 1994, including 107 000 of colon cancer and 42 000 of rectum cancer. The incidence ranks the third in males and the second in females in the United States (American Cancer Society, 1994). Epidemiological studies show that increased risk of colorectal cancer may be associated with both genetic and environmental factors (Potter et al., 1993). For the genetic component of CRC, two major types of CRC predisposition have been identified: familial adenomatous polyposis (FAP), which is believed to account for 1% of CRC cases, and hereditary non-polyposis colorectal cancer (HNPCC), which is considered to explain about 4-15% of CRC cases (Peltomaki et al., 1993 (Marks et al., 1991;Cunningham et al., 1992;Vahakangas et al., 1992;Dalbagni et al., 1993;Cordon-Cardo et al., 1994;Jacquemier et al., 1994). TP53 mutational spectra can point to particular leads in the aetiology and carcinogenesis of cancer (Harris, 1993 (Field et al., 1991;Brachman et al., 1992), oesophageal (Hollstein et al., 1991) and bladder cancers (Spruck et al., 1993;Zhang et al., 1994a). It has also been related to the ageing in prostatic adenocarcinoma (Zhang et al., 1994b) and associated with exposures to ultraviolet light in squamous cell carcinoma (Brash et al., 1991), and to aflatoxin exposure and hepatitis B virus (Hsu et al., 1993) in hepatocellular carcinomas...