The relative autonomy of the interpreted text of professional interpreters in relation to the source language text has been highlighted by various authors. Robin Setton, for example, observes that “interpreters formulate largely independently of input sentence structure, and that the contents of the last utterance is not the only source of their ongoing production” (Setton 1999: 173). This paper attempts to examine some of the reasons for this autonomy of the interpreted text of professional interpreters, especially when compared to the performance of trainee interpreters, on the basis of literature in the first part and a small corpus provided by professional and trainee interpreters working from Italian into English in the second part. The third part of the paper looks at the relevance of the findings to the teaching of interpretation.