This paper presents the analysis of three interpreter-mediated medical encounters. The aim of this study is to see whether the interpreter prioritises one participant or one conversational axis when working in a multi-party encounter and whether a common participation model is adopted by the interpreters at work. The two categories of zero renditions and non-renditions as defined by Wadensjö (1998) were used to gain an insight about what the interpreters choose not to render and what conversational initiatives are autonomously undertaken by the interpreters. The main findings of this analysis are briefly discussed and illustrated by examples.