This doctoral thesis concerns the development of methods, algorithms and a prototype of a software environment for solving multi-objective stochastic simulation-based optimisation problems with constraints and mixed decision variables. The developed simulation-based hybrid optimisation method combines the two-phase search with the compromise programming method. The two-phase search method integrates multi-objective simulation-based genetic algorithm and response surface-based linear search algorithm, which are global and local search algorithms, respectively. A global search is applied to discrete and continuous decision variables at the first phase, whereas local search works only with continuous decision variables at the second phase. The integration of global and local searches of Pareto-solutions allows achieving the diversity of solutions along the current approximation of the Pareto-optimal front, on the one hand, and preventing non-dominated solutions from being lost, on the other hand. The compromise programming method is used to select a single solution for the implementation in practice. The morphological analysis has been carried out to determine the link between the parameters of existing hybrid multi-objective evolutionary algorithms and the requirements imposed on the mutli-objective simulation optimisation of complex processes. A special focus has been given to the improvement of multi-objective simulation-based genetic algorithm by modifying an encoding mechanism and population initialisation mechanism, entering new termination criterion, additional heuristics and constraint handling technique. Approbation of the developed methods and algorithms was carried out by solving multi-echelon cyclic planning problem. It has been proved that the developed methods and algorithms provide a fast convergence and essentially reduce the number of optimisation iterations. A prototype software environment for simulation-based optimisation has been developed using ServiceModel Professional 7.0 simulation software, ProModel ActiveX Automation capability and Visual Basic for Applications programming language. The doctoral thesis contains 156 pages, 14 tables, 65 figures and 4 appendixes.