The role of obesity in the risk of colorectal cancer is uncertain. We investigated the association between height and weight and the risk of colorectal cancer in an 18-year follow-up of 5249 employed men aged 40-59 years (mean, 48 years). Cancer of the colon was diagnosed in 51 men, and cancer of the rectum in 42 (all were adenocarcinomas). Adjusted for weight and age, the tertile of men with shortest height had a relative risk (95% confidence limits) of rectum cancer of 3.1 (1.0-9.0, p = 0.04), compared with the tallest tertile. Compared with the tertile of men who weighed the most, the tertile of men who weighed the least had an increased risk of 2.5 (0.9-6.9, p = 0.08) after adjustment for age and height. Compared with men who were in the highest tertile of both height and weight, the men in the lowest tertile of both height and weight had and increased risk of 5.5 (1.2-24.9, p = 0.02). There were no significant differences in height and weight between colon cancer cases and non-cases, but colon cancer cases had a significantly lower body mass index (kg/m2), 24.4 versus 25.3 (p = 0.03). Potentially confounding factors, such as smoking, alcohol, coffee consumption, physical activity on the job and in leisure time, and social class, had no influence on the results. We conclude that low height and low weight were strong predictors of rectal cancer, and that the least obese men had the highest risk of colon cancer.