A method for predicting emissions from an emissions source. Test values of process variables relating to operation of the emissions source are gathered, along with corresponding time-correlated test values of the emissions variable to be predicted. Using the test values of the process variables, test values of a plurality of first coefficients are calculated for each process variable and associated with the process variable, and test values of a plurality of second coefficients are calculated for each value of each process variable and associated with the value of the process variable. Comparison values of the process variables relating to operation of the emissions source are gathered, along with corresponding time-correlated comparison values of the emissions variable to be predicted. Using the comparison values of the process variables, comparison values of a plurality of first coefficients are calculated for each process variable and associated with the process variable, and comparison values of a plurality of second coefficients are calculated for each value of each process variable and associated with the value of the process variable. Predetermined combinations of the comparison values of the variables and their associated coefficients are then iteratively compared with the test values of the respective variables and associated coefficients. Where the comparison yields matches between the comparison values and test values of the variables and their associated coefficients, the test values of the emissions variable associated with the matched test values of the variables are averaged and assigned as a predicted value of the emissions variable.