The records of a cohort of 11,580 females and 11,366 males participating in an Icelandic cardiovascular risk factor study were linked with the Icelandic Cancer Registry, identifying 1,785 males and 1,490 females who had been registered with neoplastic diseases from 1968 to 1995. The interval between the time of measurement of the variables and the diagnosis of the malignancy ranged from 4 to 27 years. The variables consisted of answers from a questionnaire on smoking and the use of hypertensive drugs and anthropometric and biochemical measurements. Cox's regression was applied to analyze the predictive power of the variables on the risk of cancer after the first examination at the Heart Preventive Clinic, Reykjavík. Univariate analyses, adjusted for age, were performed for each variable and each major site. Within each major site, multivariate regression analysis was applied for variables that were found significantly (10% level in univariate analysis) positive or negative as risk factors. The results show that smoking is the most important risk factor, negative only for endometrium. For lung cancer, the risk is twice as strong for females as it is for males, whereas for pancreas, males have a relative risk ratio of 4.5, compared with 2.4 for females. Height is a risk factor for all sites for each sex, for breast in females, and for kidney in males. Several anthropometric risk factors were studied. Some of these can describe positive or negative relative risk ratios for cancer, and their use may shed light on cancer pathogenesis. Serum cholesterol is a negative risk factor for breast cancer in females, but triglycerides are a positive risk factor for cervix cancer in females and for colon or rectum and thyroid cancer in males. Serum glucose is a positive risk factor for prostate cancer and a negative risk factor for lymphomas and leukemias.