The problem of identifying faulty processors (or units) in diagnosable systems is considered. For the purpose of diagnosis, a system composed of interconnected independent heterogeneous processors is modeled using a comparison graph, where tasks are assigned to pairs of processors and the results are compared. The agreements and disagreements among the units are the basis for identifying faulty processors. It is assumed that at most t processors can fail at the same time and that faults are permanent. In this paper, we introduce a new artificial-immune-based diagnosis approach using the comparison approach. The new approach has been implemented and evaluated using randomly generated diagnosable systems. Simulations results indicate that the immune approach is a viable addition to present diagnosis problems.