Pex monitors the execution of .NET application using the CLR profiling API. It uses this API mainly to rewrite MSIL instructions of selected methods, types or assemblies on-the-fly at JIT time. It inserts callbacks into the instruction sequence, which allow the observation of the execution of the program at a fine grained level.

This monitoring framework allows Pex to analyze existing test suites. Besides computing code coverage information, Pex can also dynamically check for violations of certain programming rules, e.g. whether some resources were not released. If tests fail, Pex can also often determine the root cause of the failures. Pex will suggest code snippets to the user, which will prevent the same failure from happening again.

The main component of Pex is the test case generation, which executes the code under test repeatedly in a feedback loop.