Pattern recognition methods were applied to SIMS spectra of aluminum and manganese borides to test if proper feature selection and distinction between the various compounds on the basis of the evaluation of cluster ions is possible. It was found that the features selected on an empirical basis could be reduced for practical application from 13 and 11 to 2–3 and 5–8 for aluminum borides and manganese borides, respectively, and that identification of the different compounds is possible by evaluation of the fragmentation pattern. Furthermore, there is evidence that this approach does not only work for pure compounds but also for specimens, where the interesting species is only a minor fraction in the analysed volume.