Mixed ruminal bacteria (MRB) from cattle fed hay produced ammonia from protein hydrolysate twice as fast as MRB from cattle fed mostly grain, and a mathematical model indicated that cattle fed hay had approximately four-fold more hyper ammonia-producing ruminal bacteria (HAB). HAB had a high maximum velocity of ammonia production (V(max)) and low substrate affinity (high K(m)), but simulations indicated that only large changes in V(max) or K(m) would cause a large deviation in HAB numbers. Some carbohydrate-fermenting ruminal bacteria produced ammonia at a slow rate (CB-LA), but many of the isolates had almost no activity (CB-NA). The model indicated that the ratio of CB-LA to CB-NA had little impact on HAB numbers. Validations based on predicted ratios of HAB, CB-LA and CB-NA over-predicted the specific activity of ammonia production by MRB, but co-culture incubations indicated that washed MRB from cattle fed grain could inhibit HAB. Because autoclaved MRB had virtually no effect on HAB and the incubations were always carried out at pH 7.0, the inhibition was not simply a chemical effect (e.g. low pH).