The effect of dietary habits on colorectal cancer (CRC) risk may be modified by a family history of CRC. We analyzed data from an Italian case-control study, including 1953 CRC cases and 4154 controls. Odds ratios (OR) and 95% confidence intervals (CI) for combined categories of family history and tertiles of two a posteriori dietary patterns were derived using multiple logistic regression models. Compared with individuals without family history and in the lowest tertile category of the 'starch-rich' pattern, the ORs of CRC were 1.38 (95% CI: 1.19-1.61) for the group without family history and in the highest tertile, 2.89 (95% CI: 2.30-3.64) for the one with family history and in the lowest tertile, and 4.00 (95% CI: 3.03-5.27) for the one with family history and in the highest tertile. Compared with individuals without family history and in the highest tertile of the 'vitamins and fiber' pattern, the ORs were 1.29 (95% CI: 1.12-1.48) for the group without family history and in the lowest tertile, 2.89 (95% CI: 2.30-3.64) for the one with family history and in the highest tertile, and 3.74 (95% CI: 2.85-4.91) for the one with family history and in the lowest tertile. Family history of CRC and 'starch-rich' or 'vitamins and fiber' patterns has an independent effect on CRC risk in our population. However, as having a family history plausibly implies shared environmental and/or genetic risk factors, our results could not exclude that dietary habits can modify genetic susceptibility to CRC.