The process of stereo vision matches one or more images to recover the depth information of the pictured scene. Great progress is currently being made within this field, with algorithmic research, computational power developments, and cheaper cameras all contributing to give stereo vision great future potential as the depth measuring system of choice. One of the challenges of the stereo vision approach is the multitude of control parameters, which all affect algorithm behaviour. These parameters have traditionally been tuned by hand, with some limited use of computerized optimisation techniques. However, the process of evolutionary computation provides a promising method of optimisation of such complex problems. This thesis explored the possibility of applying a multi-objective evolutionary optimisation approach to the stereo algorithm parameter problem. In this regard, an automatic parameter optimisation framework based on the multi-objective optimisation algorithm NSGA-II was developed and tested. In order to judge the performance of the framework and the validity of the multi-objective approach, three different stereo algorithms were tested and a series of near pareto-optimal parameter sets were produced. One parameter set per algorithm was submitted to the official KITTI stereo vision benchmark ranking, and was able to improve upon the current official results.