This work presents a procedure to optimize the molecular geometry at the Hartree-Fock level, based on a global opti-mization method—the Generalized Simulated Annealing. The main characteristic of this methodology is that, at least in principle, it enables the mapping of the energy hypersurface as to guarantee the achievement of the absolute minimum. This method does not use expansions of the energy, nor of its derivates, in terms of the conformation variables. Distinctly, it performs a direct optimization of the total Hartree-Fock energy through a stochastic strategy. The algorithm was tested by determining the Hartree-Fock ground state and optimum geometries of the H2, LiH, BH, Li2, CH+, OH?, FH, CO, CH, NH, OH and O2 systems. The convergence of our algorithm is totally independent of the initial point and do not require any previous specification of the orbital occupancies.