A methodology for simulating harmonic regressions is presented that allows for the stochastic simulation of the harmonic regressions when various orders of autocorrelation are present. Statistical properties of the historical correlations are respectably maintained in an empirical example using ten livestock classes which exhibit first and twelfth order autocorrelation.