Eighty-four patients with a total of 104 admissions for intestinal obstruction who each also had a history of cancer had their records reviewed to determine what variables, if any, would help predict outcome. Multivariate discriminate analysis was used to assign the patient into one of three outcome possibilities: 1) alive and well, 2) alive with intestinal obstruction, and 3) dead. The computer accurately assigned outcome 71 per cent of the time. Twenty-four per cent of the patients had no cancer found at laparotomy, and had good results. Patients with carcinomatosis did poorly. Females fared much better than males. The natural history of patients with intestinal obstruction and cancer is that about 35 per cent leave the hospital eating normally, 20 per cent are alive but unable to eat, and 45 per cent die on the same hospital admission or shortly thereafter.