Fault injection is become a popular approach to evaluate and possibly to improve the dependability of computer-based systems. One of the main issues to be solved when setting up a fault injection experiment is the generation of a list of faults to be injected, really representative of the whole set of possible faults. This paper proposes a set of collapsing rules based on the analysis of the assembly code and of the behavior of a fault free run of the system, useful to reduce the fault list length and the fault injection time without decreasing the accuracy of the results. The approach is suitable to be adapted for microprocessor-based systems and is independent on the method used to generate the fault list to be collapsed.