In this study, analysis of a multi-response experimental data set was considered. The multi-response experimental data set, which have linearly correlated responses, was analyzed in three stage. In the first stage, called modeling, the multi-response experimental data set was modeled by using Seemingly Unrelated Regression (SUR) due to linear relationship between the responses. The second stage, called multi-objective optimization, was achieved by using multi-objective metaheuristic methods, Non-dominated Sorting Genetic Algorithm-II (NSGA-II) and Multi Objective Differential Evolution (MODE). In the last stage, called decision making, the obtained results were discussed and compromise experimental conditions were determined by using Multi-Criteria Decision Making (MCDM) methods, called TOPSIS, MABAC, CODAS. It is seen from the results that parameter estimates of the SUR method has more reliable than the ordinary least square (OLS) estimates. Pareto solution set of the MODE is more prefable according to the performance metrics and the TOPSIS and the CODAS give same compromise solution among the Pareto solutions. The obtained compromise solution can be used as an experiment condition by the researcher confidently.