Emissions reduction for diesel engines can be approached either directly by exploratory testing or indirectly by using a mathematical model to predict low-emission modifications. The empirical approach suffers from excessive costs and limited extrapolations from given engines; at the same time, theoretical predictions are too uncertain to use alone because of lack of understanding of the diesel combustion mechanism. The two approaches are complementary and should be pursued together. The authors have established a foundation for a mechanistic model by four activities: emissions data was generated and correlated with changes in engine parameters; existing models were critically assessed; a mechanistic model of heat release was outlined and an analysis of key questions was begun; and diesel flame measurements were conceived which can resolve modeling issues and thereby insure that model development will be cost effective.