The new control techniques of off-line model predictive control and level set methods have been applied to the challenges of engine management for diesel exhaust aftertreatment. Theoretical work has been supported by experiments carried out in a diesel engine test cell. A control relevant air-path model was developed using experimental work carried out on a diesel engine and dynamometer, with the aim of improving the engine control during the transients required by aftertreatment devices. This was considered as a two input and two output system, using an intake throttle and a valve associated with the exhaust gas recirculation system to control the in-cylinder quantities of fresh air and exhaust gas. Model predictive control techniques were used to design a hybrid controller, which was compared to a traditional proportional and integral controller and a 1-{00 controller in simulation. Level set methods were used to analyse supervisory hybrid control of a lean NOx trap model. In this system periodic mode changes are required to purge the NOx trapped on the filter and convert it to less harmful products. A novel diesel particulate filter model suitable for control was developed to represent the thermal wave behaviour observed in worst-case regenerations, where the soot in the filter burns to create local high temperature regions. A DPF was instrumented and fitted to the diesel engine and tested in the development of this model, particularly focusing upon these worst-case conditions. This new theoretical model was analysed using level set methods in a comparison with a simple control oriented DPF model. A supervisory hybrid control strategy to determine when to take emergency action to avoid thermal damage to the filter was designed using the simple DPF model. This strategy used safe conditions verified from the experimental work undertaken .