Abstarct: In some problems Applying heuristic methods are tedious and sometimes inoperative and specially in structural optimization that we deal with high scaled structures are very expensive. The unsuitable reasons these methods are, high design variables, to get non-linear functions, or implicit functions. It causes to tabernacle approximating methods while it doesn’t damage main problems. A method for structural optimization should be flexible and general. It should be able to handle not only element sizes as design variables, but also, for instance, shape variables and material orientation angles. It should be also be able to handle ‘all kinds’ of constraints, provided only that the derivatives of the constraint functions with respect to the design variables could be calculated. Thus, the method should be able to handle general non-linear programming problems. Prof. Krister Svanberg believe that these requirements and wishes are to a rather large extent met by the method of moving asymptotes (MMA) and MMA is easy to implement and use. Here, we try to explain general and approximating method in optimization specially MMA, Shortly. Then we present a program by FORTRAN by using MMA for solution of topology optimization of plane stress problems and compare its result with optimality criteria. Then we present a program with stress constraint and we show the influence of changing parameters on result. Finally we compare our program with traditional method such as SLP and approximating method such as CONLIN by changing in the code of MMA.