A notion of test for intuitionistic type theory has been recently introduced by Peter Dybjer. It is meant to be the foundation for automatic testing tools that could be implemented in proof assistants such as Coq or Agda. Such tools would provide a way to test, at any time during the construction of a proof, if the current goal is typable in the context. The failure of such a test would mean that the goal is impossible to prove, and its success would corroborate the partial result. We investigate the possibility of extending the testing procedure to classical systems, and propose an interpretation of the testing procedure in term of Krivine's classical realizability.