Modern advanced control methods have been identified as one key step to improve emission reduction as well as increased reliability and stability. Current control loops on legacy gas engines are acting mainly in independent cascaded PID loops. As a result, these separate controllers tend to cause an unstable behavior during transient changes. This can lead to significant emission excursions requiring large compliance margins when operating at low emissions levels. An alternative approach is to replace the independent loops by a state-based controller which simultaneously sets all the loops. A Model Predictive Control (MPC) for two stroke gas engines has been developed as part of the ERLE program (Emission Reduction for Legacy Engines) initiated by the Pipeline Research Council International (PRCI). A dynamic gas engine model suitable for prediction of emission and engine behavior serves as the core of this control approach. The paper will describe the development of the process model for a TLA 6 using real time transient measurements and advanced simulation. This was in turn used to develop an explicit predictive control system which was then implemented in a fast prototyping system and tested on the TLA 6. The results of the initial field test together with a benefit assessment and further development steps will be presented.