The minimum cost perfect matching problem is one of the rare combinatorial optimization problems for which polynomial time algorithms exist. Matching algorithms find applications in Postman Problem, Planar Multicommodity Flow Problem, in heuristics to the well known Traveling Salesman Problem, Vehicle Scheduling Problem, Graph Partitioning Problem, Set Partitioning Problem, in VLSI, et cetera. In this thesis, reviewing the existing primal-dual approaches in the literature, we present two efficient algorithms for the minimum cost perfect matching problem on general graphs. In both of the algorithms, we achieved drastic reductions in the total number of time consuming operations such as scanning, updating dual variables and reduced costs. Detailed computational analysis on randomly generated graphs has shown the proposed algorithms to be several times faster than other algorithms in the literature. Hence, we conjecture that employment of the new algorithms in the solution methods of above stated important problems would speed them up significantly.