Eighty four colorectal cancer patients who underwent presumably curative surgery were considered as candidates for control recurrence study. The study protocol included physical exam, blood biochemistry, chest X-ray, and CEA every 3 months (every 6 months after the third year) as well as an opaque enema, colonoscopy, abdominal echography, and pelvic CT scan (in case of abdominoperineal resection of a distal rectum cancer) every six months (once a year after the third year). Thirteen patients (15%) refused to participate in the study or abandoned it early, thus 71 patients are finally evaluated. After a mean follow up of 64 months, 32 patients presented a total of 42 recurrences (several patients underwent complete resection of the recurrence and presented new relapses). Recurrence diagnosis was always done by conventional exam (there was no second look guided by CEA values). Complete resection of the initial recurrence was achieved in 22% of cases (7 patients), but only 2 (6.5% of total relapses) are up to date tumor free and potentially cured after a long follow up. The most useful tests for the diagnosis of organ specific recurrences were chest X-ray (lung metastasis), liver echography (liver metastasis), liver echography (liver metastasis), colonoscopy (intraluminal local recurrence) and pelvis CT scan (pelvic recurrences). The carcinoembryogenic antigen was very specific, but only moderately sensitive, in predicting tumoral recurrency. However, CEA levels at relapse predicted the complete resection of recurrent tumor (58% with carcinoembryogenic antigen equal to or lower than 10 ng/ml compared to 5.5% with carcinoembryogenic antigen above this value p less than 0.001).(ABSTRACT TRUNCATED AT 250 WORDS)