In the present study, an efficient combination of the Tau method with the Bernoulli polynomials is proposed for computing the Feedback Nash equilibrium in differential games over a finite horizon. By this approach, the system of Hamilton-Jacobi-Bellman equations of a differential game derived from Bellman's optimality principle is transferred to a nonlinear system of algebraic equations solvable by using Newton's iteration method. Some illustrative examples are provided to show the accuracy and efficiency of the proposed numerical method.