We conducted a population-based case-control study among different ethnic groups in Hawaii to evaluate the role of various types and components of fiber, as well as micronutrients and foods of plant origin, on the risk of colorectal cancer. We administered personal interviews to 698 male and 494 female Japanese, Caucasian, Filipino, Hawaiian, and Chinese cases diagnosed during 1987-1991 with adenocarcinoma of the colon or rectum and to 1,192 population controls matched to cases by age, sex, and ethnicity. We used conditional logistic regression to estimate odds ratios, adjusted for caloric intake and other covariates. We found a strong, dose-dependent, inverse association in both sexes with fiber intake measured as crude fiber, dietary fiber, or nonstarch polysaccharides. We found inverse associations of similar magnitude for the soluble and insoluble fiber fractions and for cellulose and noncellulosic polysaccharides. This protective effect of fiber was limited to fiber from vegetable sources, with an odds ratio of 0.6 (95% confidence interval = 0.4-0.9) and 0.5 (95% confidence interval = 0.3-0.7) for the highest compared with the lowest quartile of intake for men and women, respectively. We found associations of the same magnitude for soluble and insoluble vegetable fiber, but no clear association with fiber from fruits or cereals. This pattern was consistent between sexes, across segments of the large bowel (right colon, left colon, and rectum), and among most ethnic groups. The effect of vegetable fiber may be independent of the effects of other phytochemicals, since the effect estimates remained unchanged after further adjustment for other nutrients. Intakes of carotenoids, light green vegetables, yellow-orange vegetables, broccoli, corn, carrots, bananas, garlic, and legumes (including soy products) were inversely associated with risk, even after adjustment for vegetable fiber. The data support a protective role of fiber from vegetables against colorectal cancer, which appears independent of its water solubility property and of the effects of other phytochemicals. The data also indicate that certain vegetables and fruits may be protective against this disease through mechanisms other than their fiber content.