There were many studies in the last fifty years observing errors in simultaneous interpreting, (Altman 1994, Barik 1994, Gerver 2002, Pöchhacker 2004, among many). Their main focus was error analysis in simultaneous interpreting by professional interpreters, with the exception of Altman’s pilot-study (1994) examining errors that occurred in simultaneous interpreting by trainee interpreters. However, none of them dealt with students’ progress during a particular time span. This paper analyses error types and their frequency in simultaneous interpreting by trainee simultaneous interpreters, which include errors on the level of lexemes, collocations and phraseological units. The analysis examines the first simultaneous interpretation, recorded before the training occurred and the final one, recorded after four months of training. The relationship between the source text and the target text in particular moment is observed and determined, as well as the increase or decrease of certain types of errors during four months of training. The aim of this paper is to determine the most frequent error types committed by trainee simultaneous interpreters, classification of errors on the basis of Barik’s (1994) coding scheme, to ascertain which error types occur less frequently in the final interpretation and to try to identify possible reasons for persistence of those errors that show similar level of occurrence in both interpretations. The research included 10 second-year graduate students of translation studies. An average of 5 minutes of simultaneous interpretation of the first and final speech was analyzed, so the corpus includes 100 minutes of simultaneous interpreting