A hybrid algorithm based conjugate direction method and modified simulated annealing algorithm for mobile robot global path planning is proposed in this paper. On each temperature, the hybrid algorithm firstly uses conjugate direction method to search local optimal solution, then uses modified simulated annealing algorithm to move off the local optimal solution, and then the temperature is updated. These operations are repeated until to obtain global optimal solution. The computer simulation results demonstrate, that this hybrid algorithm is feasibility, and better than simulated annealing algorithm on optimization, and its convergence speed is quicker than that of simulated annealing algorithm.