The field of simultaneous interpretation (SI), and more specifically that of the methodology of teaching SI, will always welcome new research. Hence, it is to be hoped that an analysis of the most frequently recurring types of error might make it possible to predict, and perhaps even forestall, the production of errors by student-interpreters, thereby contributing to improvements in teaching methods. The comments presented in this paper are limited in scope, since they are based upon the observation of one investigator alone and on a restricted sample of empirical data. Perhaps they might serve to suggest a method whereby a more exhaustive study could be undertaken.