There exist many studies conducted to compare the performance of different production control mechanisms (PCMs) in order to determine which one performs the best under different conditions. Nonetheless, most of these studies suffer from the problems that the PCMs are not compared with their optimal parameter settings in a truly multi-objective context. This paper describes how different PCMs can be compared under their optimal settings through generating the Pareto-optimal frontiers, in the form of optimal trade-off curves in the performance space, by applying evolutionary multi-objective optimisation to simulation models. This concept is illustrated with a bi-objective comparative study of the four most popular PCMs in the literature, namely Push, Kanban, CONWIP and DBR, on an unbalanced serial flow line in which both control parameters and buffer capacities are to be optimised. Additionally, it introduces the use of normalised hyper-volume as the quantitative metric and confidence-based significant dominance as the statistical analysis method to verify the differences of the PCMs in the performance space. While the results from this unbalanced flow line cannot be generalised, it indicates clearly that a PCM may be preferable in certain regions of the performance space, but not others, which supports the argument that PCM comparative studies have to be performed within a Pareto-based multi-objective context.