Abstract: Regarding the saddle-po int equ ilibrium for discrete t ime bilinear-quadratic controlin the f in ite t ime, the sadd le-po int equ ilibrium is converted into a non linear tw o po int boundary va lue by using the dynam ic prog ramm ing pr incipa.l Then by using a transfo rmation, the nonlinear tw o po int boundary value is transfo rmed in to a "linear" two point boundary va luew ith "separate" form. Finally, a new iterative a lgorithm is constructed to solve the problem, w hich provides a new approach to the problem w ith the differential gam es o f discrete-time b ilinear systems.