Canine high-risk surgical patients were identified, and the variables that distinguished between survivors and nonsurvivors were determined. Fifty-two nonsurvivors and 176 survivors were randomly drawn from a population of 615 dogs undergoing general surgical procedures and entering the intensive care unit. Forty-two variables were examined to determine which variables were significantly different between nonsurvivors and survivors. Properative variables that were significantly different were age, hospital days prior to surgery, the number of concurrent diseases, American Society of Anesthesiologists (ASA) status, PCV, TP, platelet count, albumin concentration, alkaline phosphatase activity, bilirubin concentration, and surgery site (laparotomy = highest risk). Age, PCV, TP, platelet count, albumin concentration, alkaline phosphatase activity, and bilirubin concentration were then examined, using stepwise logistic regression for their ability to predict mortality in a separate population of 169 canine patients undergoing laparotomy. Age, TP, and platelet count were identified as predictive variables.