Evidence is accumulating that individual risk of neoplasia depends on complex interactions among genetic inheritance, a range of exposures both in utero and in postnatal life, and the play of chance. Knowledge of the portfolio of genetic variants that confer susceptibility or resistance to cancer is limited, and there is potential for genome-wide scans and hypothesis-driven studies to reveal novel polymorphisms and haplotypes that modify risk. There is only fragmentary evidence of the scale and nature of diet-gene interactions that modulate risk of neoplasia, but it seems probable that such interactions will play a significant role as they do in other complex diseases including cardiovascular disease and type 2 diabetes. All existing evidence about diet-gene interactions and cancer risk comes from observational studies, and it will be necessary to undertake intervention studies to test the hypotheses generated by epidemiologic investigations. Because it is very unlikely that primary cancer will be an endpoint in dietary intervention studies in the foreseeable future, development of robust surrogate endpoints is a high priority. Emerging biological science using epigenomics, proteomics, and other molecular technologies appears to offer novel approaches to the discovery and validation of surrogate endpoints.