We evaluated the relation of specific sources and components of fiber with diagnosis of distal colon (n = 531) or rectal (n = 159) adenomatous polyps or hyperplastic (n = 327) polyps in the Health Professionals Follow-up Study. We studied 16,448 men free of cancer or polyps in 1986, who underwent endoscopy in 1986-1994, and who provided diet and medical history. Relative risks (RRs) and 95% confidence intervals (CIs) adjusted using multiple logistic regression were calculated. We observed a modest reduced risk of distal colon adenoma with increasing intake of fiber from fruit (P-trend = 0.03) but not cereals or vegetables. The RR comparing the highest (median, 8.4 g/day) to lowest (1.3 g/day) quintile of fruit fiber intake was 0.81 (95% CI, 0.59-1.11). Soluble fiber, but not insoluble fiber, appeared to be inversely associated with distal colon adenoma (P-trend = 0.007). Comparing extreme quintiles (9.4 versus 3.4 g/day soluble fiber), the RR was 0.69 (95% CI, 0.46-1.03). Polyps detected in 1986 or later among men also with a negative endoscopy before 1986 may be considered to be "incident," with diet report corresponding more closely to time of polyp development. For "incident" cases (n = 130), the relation between soluble fiber and distal colon adenoma was strengthened (extreme quintiles RR, 0.27; 95% CI, 0.11-0.66; P-trend = 0.003), whereas for "prevalent" cases (n = 401), we found no association. No consistent relation between fiber and rectal adenomas or hyperplastic polyps was observed. These results suggest that soluble fiber may be particularly important in reducing risk of adenomatous polyps of the distal colon and support national dietary guidelines of increasing fruit consumption.