An experiment was conducted to test the accuracy of the metabolizable protein system in predicting the amount of urea that would be useful in a corn-based cattle diet. Treatment diets included a basal, low-protein (7.8% CP) negative control (NC) with no supplemental N and a positive control (PC) that contained soybean meal. Urea was added to the NC diet in quantities calculated to be either 25% deficient (LU), equal to (MU) or 25% in excess (HU) of the urea fermentation potential ( UFP ). In vitro rumen fermentation studies were used to determine sequential ammonia production and digestible dry matter content of the diets. In a growth trial, 12 individually-fed Angus, Hereford and Angus X Hereford steers weighing an average of 213 kg were assigned randomly to each treatment diet. At the conclusion of the 112-d trial, rumen ammonia and jugular blood urea N (BUN) concentrations were determined on two steers from each treatment before feeding and at 1, 2, 3, 4 and 5 h postfeeding. In vitro ammonia concentrations of the NC and PC treatments were lower (P less than .05) than that of urea containing diets. In vivo rumen ammonia concentrations at 1 h postfeeding and BUN levels at 3 h postfeeding were low for both the NC and PC diets compared with urea-containing diets. Both of these values increased with each successive increase of added urea to the NC diet. During the initial 70 d of the growth trial, daily gains were improved (P less than .05) by addition of urea up to the MU level, which fulfilled the calculated UFP .(ABSTRACT TRUNCATED AT 250 WORDS)