We conducted a prospective study using primary colonoscopy in the families of patients with surgically treated colorectal cancer (CRC) to evaluate its feasibility as a screening method for CRC. The presumed detection sensitivity of sigmoidoscopy in this group was also evaluated. Over a 3-year period, 142 first-degree relatives (73 men and 69 women) of 117 patients with CRC participated in this colonoscopic screening study. The average age at presentation for examination was 49.1 years (range, 31-79 yr). Forty-nine patients had adenomatous polyps (AP) and four patients had cancer, yielding a frequency of 37% (53/142). Thirty-six family members were under 40 years of age, eight of whom had AP. In total, there were 111 AP and four invasive cancers; 42% (48/111) of AP and 75% (3/4) of cancers were beyond the reach of conventional flexible sigmoidoscope. Of 53 patients with AP or cancer in the large bowel, 36 had AP or cancer in the distal colon (rectosigmoid region and descending colon), 15 (42%) of whom were found to have at least one AP or cancer in the proximal colon (proximal to the descending colon). The remaining 17 patients' lesions were confined to the proximal colon, and would not have been detected with a flexible sigmoidoscope. Sigmoidoscopy in high-risk patients with neoplasm in the large bowel has a sensitivity of 68%. We conclude that a vigorous surveillance program using primary colonoscopy should be set up for those who are first-degree relatives of patients with CRC. Routine examination should start at the age of 40, and the entire colon should be evaluated.