Twin studies integrate genetic and environmental (including physical environment and life-style) information by comparing monozygotic and dizygotic twins for the occurrence of disease. Our objectives were to compare cancer incidence in twins with national rates and to estimate both the probability that co-twins of affected twins may develop cancer and the importance of genetic predisposition and environment in cancer development. The nationwide record linkage of the Finnish Twin Cohort Study, the Finnish Cancer Registry and the Central Population Register allowed the follow-up of 12,941 same-sexed twin pairs for incident primary cancers from 1976 to 1995. Zygosity was determined by use of a validated questionnaire in 1975. Methods included calculation of standardized incidence ratios and concordances and fitting of structural equation models. A total of 1,613 malignant neoplasms occurred in the cohort. The overall cancer incidence among twins resembled that among the general population. Monozygotic co-twins of affected twins were at 50% higher risk than were dizygotic co-twins. Based on genetic modeling, inherited genetic factors accounted for 18% (95% confidence interval 4-32%) of the liability in inter-individual variation in the risk of overall cancer, while non-genetic factors shared by twins accounted for 7% (0-16%) and unique environmental factors for 75% (65-85%). Our results appear to exclude a contribution greater than one-third for genetic predisposition in the development of cancer in the general population, thus pointing to the earlier confirmed substantial role of environment.