The relationship between diet and subsite-specific colon cancer was investigated using dietary histories obtained from a statewide, population-based sample of 152 proximal colon cancer patients, 201 distal colon cancer patients and 618 general population controls. The results do not support hypotheses that (1) dietary fat and cholesterol are more strongly related to proximal colon cancer and (2) vegetables and other dietary sources of fiber are more strongly associated with distal colon cancer. Vegetable consumption over lifetime was consistently protective for both proximal and distal colon cancer. Odds ratios and 95% confidence intervals for the most significant dietary factors (based on high vs. low consumption) for proximal colon cancer were: salad, 0.29 (0.17, 0.48); miscellaneous vegetables, 0.58 (0.35, 0.97); cruciferous vegetables, 0.59 (0.35, 0.97); processed lunchmeat, 2.04 (1.31, 3.17); pan-fried foods, 1.79 (1.15, 2.80); eggs, 1.75 (1.02, 2.99) and for distal colon cancer they were: salad, 0.43 (0.28, 0.67); cruciferous vegetables, 0.44 (0.28, 0.71); cheese, 0.62 (0.40, 0.96); processed lunchmeat, 1.79 (1.17, 2.73); pan-fried foods, 1.55 (1.03, 1.27). The results support recommendations that the "prudent diet" (low-fat, high-vegetable) may reduce colon cancer risk.