Due to stringent regulations, tremendous efforts have been made worldwide in monitoring and diagnosing combustion engines. However, there is still a large need to further develop optimization models that facilitate the understanding of the relationship between engine parameters and parameters such as noise and exhaust emissions. The aim of present study was to investigate the potential use of vibration based diagnostics for prediction of different parameters such as noise, exhaust emissions, Pmax and the dp/d alpha . The method is based on a reconstruction of the cylinder pressure from vibration measurements on the engine surface. A three factorial central composite face design was used for the tests involving different running conditions (i.e. speeds and loads) and different blends of rap seed oil methyl esther (RME)/ethanol. Principal component analysis (PCA) and partial least squares (PLS) were thereafter used for establishing models that show the relationship between speed, load, amount of RME and responses such as cylinder pressure, exhaust emissions and sound pressure. The results show that the reconstructed cylinder pressure can be used for diagnostics and control by allowing an accurate estimation of Pmax and dp/d alpha . Furthermore, the method used is also applicable for determining apparent net heat release rate and hence the exact time of the start of the combustion process. A comparison between measured and predicted values of NOx, noise, Pmax and dp/d alpha showed a good predictive power of the established models.