The engine manufacturers have to meet more and more stringent regulations on both noise and exhaust emissions in the future requiring extensive optimisation of all parameters of the engine. Soft modelling with statistically designed experiment (SDE) and Multivariate Analysis (MVA) can be an aid in this work. The analyses are based on cylinder pressure data, engine data and exhaust emissions data, with all measurements performed simultaneously. The effects of changes in fuel quality, load, speed, and injection timing on combustion noise and exhaust emissions were investigated using SDE and MVA. The engine was a modified 6 cylinder, direct injection diesel engine running on ethanol. The paper demonstrates the applicability of experimental design and multivariate methods for modelling and minimisation of noise and exhaust emissions.