Conjugate gradient (CG) algorithms are an optimization method with fast convergence. To date, several CG methods have been developed to improve computational performance and have been used to solve unconstrained optimization problems. In this paper, a new CG algorithm based on the modified secant condition is proposed to solve unconstrained optimization problems. The proposed algorithm has the following properties:it achieves sufficient descent property and global convergence property, We used a number of test functions for large-scale unconstrained optimization problems and the numbers showed that the proposed algorithm works better than other algorithms like Hestenes and Steifel (HS) and Fletcher and Revees (FR).