Eighty-nine admissions into Auckland Hospital with exacerbation of ulcerative colitis (between May 1985 and May 1991) were analyzed to determine the rate and any predictors of medical treatment failure. Twenty-five patients (28%) failed to respond to medical treatment and required surgery. Clinical information and laboratory indices available within 24 h of the admissions were compared between the patients requiring surgery and those who did not. There were no significant differences between the groups in sex distribution, age of onset of disease, duration of disease, hemoglobin concentration, or white cell count. However, "severe" diarrhea (chi 2 = 24.83, p = 0.0001), and lower albumin level (F1, 83 = 45.61, p = 0.0001) were noted in the surgical group. There was a tendency toward higher ESR (F1, 82 = 3.06, p = 0.08) and "extensive" colitis (chi 2 = 3.29, p = 0.07) in the patients requiring surgery. Univariate analysis confirmed albumin level and severity of diarrhea as significant discriminators. A discriminant function analysis showed that albumin level and severity of diarrhea would distinguish between surgical and nonsurgical outcome in 82% of cases. Distal colitis and mild to moderate diarrhea had negative predictive values of 80% and 91%, respectively, for nonsurgical outcome of acute ulcerative colitis. It is concluded that the above significant variables are good predictors of outcome of medical treatment for exacerbations of ulcerative colitis and that the proportion of patients needing surgery has not changed in the last 35 yr despite various management strategies.