A system for evaluation of cheese yield, fat recovery, and compositional control for use in large cheddar cheese factories was developed. The manufacturing performance system was evaluated in large commercial factories and inaccurate cheese composition input data, produced by near-infrared (NIR) analysis and the coulometric method, was identified as a weakness that limited the accuracy of the performance evaluation. A new cheese analysis method using mid-infrared (MIR) analysis was developed. The MIR produced more accurate results than the NIR during a validation study run with four cheese factories. Standard error of prediction (SEP) values for moisture and fat on the NIR were 0.30 and 0.45, respectively, whereas the MIR produced SEP values of 0.28 and 0.23 for moisture and fat, respectively. MIR also out-performed the coulometric method for salt determination with SEP values of 0.036 and 0.139, respectively. The MIR had an SEP value of 0.19 for estimation of protein.