Smart card are often the target of software or hardware attacks. The most recent attack is based on fault injection which modifies the behavior of the application. We propose an evaluation of the effect of the propagation and the generation of hostile application inside the card. We designed several countermeasures and models of smart cards. Then we evaluate the ability of these countermeasures to detect the faults, and the latency of the detection. In a second step we evaluate the mutant with respect to security properties in order to focus only on the dangerous mutants.