Mechanism-based modeling is an approach in which the physiological, pathological and pharmacological processes of relevance to a given problem are represented as directly as possible. This approach allows us (i) to test whether assumed hypotheses are consistent with observed behaviour, (ii) to examine the sensitivity of a system to parameter variation, (iii) to learn about processes not directly amenable to experimentation, and (iv) to predict system behavior under conditions not previously experienced. The paper illustrates different aspects of the application of mechanism-based modeling through three different examples of relevance to the treatment of diabetes and hypertension: subcutaneous absorption of insulin, pulsatile insulin secretion in normal young persons, and synchronization of the pressure and flow regulation in neighbouring nephrons. The underlying ideas are that each regulatory mechanism represents the target for intervention and that the development of new and more effective drugs must be based on a deeper understanding of the biological processes.