A Markov chain dynamic programming model is presented for determining optimal range improvement strategies as well as accompanying livestock production practices. The model specification focuses on the improved representation of rangeland dynamics and livestock response under alternative range conditions. The model is applied to range management decision making in the Cross Timbers Region of central Oklahoma. Results indicate that tebuthiuron treatments are economically feasible over the range of treatment costs evaluated. Optimal utilization of forage production following a treatment requires the conjunctive employment of prescribed burning and variable stocking rates over the treatment's life.