The advantages and disadvantages of heterogeneous controllers, composed of multiple simple controllers, for regulating nonlinear processes are studied. These controllers are developed to provide more robustness in the face of sizable disturbances and faults. Qualitative reasoning techniques are applied to reason with incomplete knowledge of the plant and its disturbances. The qualitative algorithm QSIM (Kuipers, 1986) provides the framework for developing and analyzing models at different levels of abstraction, based on the available information. Together with NSIM, which permits the parameters and functions to be bounded, its effectiveness for the construction of stability proofs is demonstrated.