Dietary habits are thought to be involved as determinant of breast and colorectal cancer. Nevertheless results of epidemiological studies on diet show several inconsistencies. This is true for the findings related to energy and its sources. Between 1991 and 1996, 2569 women with incident breast cancer (median age: 55 years) and 2588 controls (median age: 56 years), and 1953 subjects with cancer of the colon-rectum (median age: 62 years) and 4154 controls (median age: 58 years) were interviewed in the hospitals of six Italian areas. The validated food frequency questionnaire included questions on 78 foods and recipes and specific questions on individual fat intake pattern. Significant risks for breast and colorectal cancer emerged with increasing intake of energy (odds ratios in highest vs. lowest quintile were 1.32 and 1.49 respectively). Due to the high interrelations existing among the various sources of energy, the separated analysis of each macronutrient didn't achieve the independent estimates of the effects. In order to overcome this situation, we used a completely partitioned model in which all the main sources of energy were entered simultaneously as continuous variables in the regression. High intake of starch led to an increase of cancer risk (odds ratios for an addition of 100 kcal/day were 1.08 and 1.10 for breast and colorectal cancer respectively). A positive association was also found for saturated fat (odds ratios 1.16 for breast and 1.12 for colorectal cancer). High intakes of polyunsaturated fatty acids (chiefly derived from olive and seed oils) were protective more markedly for breast cancer. A possible interpretation of the risk for starch, implies the glycemic overload and hyperinsulinemia due to the high grade of refinement of cereals (the main source of starch) eaten in Italy.