Abstract: A supply chain system is a non-linear and complex system. In order to optimize the total cost of a supply chain to enhance overall competitiveness, a simulationbased optimization approach was raised. Firstly, the System Dynamics model of a three-echelon collaborative supply chain was built. In order to minimize the total cost, enterprises in this supply chain needed to solve some decision variables. Secondly, the processes about converting the system dynamics model to a Matlab program model were minutely shown. Some groups of stochastic parameters were used to verify the converting processes. Finally, an appropriate Genetic Algorithm was applied to obtain some satisfactory solutions. The results show that this simulationbased optimization approach could not only display the dynamic characteristics of a supply chain, but also solve the problem fast with a high accuracy. It could effectively solve single-objective or multi-objective programming problems in supply chains. A system dynamics model has a relatively fixed structure and does not adapt to a heterogeneous interactive environment. This approach supplements a system dynamics model. The converted Matlab program model could be deeply simulated and studied. Objective functions and constraints could be flexibly defined. Dynamic changing processes of the system could be minutely analyzed and interactive interfaces to other applications also provided.