Nowadays, milling process is one of the most widely used metal processing methods in many fields from space and aircraft to automotive industry. The surface roughness values of the workpiece in milling process vary depending on the thermal, chemical and abrasive loads that occur during cutting. Spindle speed, depth of cut and feed rate are the cutting parameters affecting the surface roughness. Hence, these parameters at the time of machining constitute an important issue. Accordingly, in this thesis optimization of surface roughness has been performed using the stochastic search methods. First, using experimental data obtained in the milling process, it was aimed to establish a regression model to determine average surface roughness equation as an objective function. The cutting parameters and average surface roughness value were considered as input and output in regression analysis, respectively. In this study, seven different mathematical models have been established and examined to carry out regression analysis. The reliability and stability of the mathematical models were investigated. The most appropriate mathematical model has been constructed and then used as an objective function for optimization. Nelder-Mead, Random-Search, Simulated Annealing, and Differential Evolution were the stochastic search algorithms to perform the optimization in the present study. In conclusion, it was found that the minimum average surface roughness value depends on spindle speed, depth of cut and feed parameters.