A key element of the Quality by Design (QbD) paradigm of drug development is a delineation of a multi-dimensional design space for material and process parameters. Mathematical models can serve as a powerful tool in this approach. A design space could potentially include a combination of mechanistic and regression models. Mechanistic models mimic the underlying physical chemical phenomenon while the inherently empirical regression models are derived from statistically designed experiments. This talk would present an overview of role of mathematical models in the current paradigm of science based development and in addition illustrate FDA's perspective about this approach.