Abstract: A routine of active set method combined with the idea of solving multiple objective programming is proposed for solving LP. Owing to the memory-compacting technique for sparse matrix, It can solve the large scale LP problems with 3000 decision variables, 1500 constraints not including upper and lover bound constraints and nonzero elements of oriqinal data is less than 15000 large scale linear programming problems except those on the right on the microcomputers such as IBM-PC and its compatible computers with 512K memory. It is much more convenient than the simplex method in inputing data, modifing models and particularly in adding or deleting constraints. Its execution speed is also faster additionally.