Strongly positive correlations exist between the incidence of breast and large bowel cancers in different countries and the corresponding per caput intake of fat. However, attempts to confirm these relationships have, on the whole, proved disappointing, in studies of individuals whether using the case-control or the prospective approach. Moreover no positive relationship is evident in prospective studies between blood lipid levels and colon or breast cancers. Overall the human evidence for fat as a cause of these cancers is weak. Possible reasons are considered for the findings in different studies.