In oral proficiency tests there are occurrences of trouble in interaction such as misunderstanding, non hearing or lack of understanding which may cause breakdown in communication. Within the context of the question answer framework of an oral proficiency test this thesis investigates the interactive nature of spoken discourse between students and assessors when there is trouble in talk as perceived by the assessors, with a focus on how they accommodate to the students. A sample of 20 oral transcripts of the 1992 Victorian Certificate of Education Italian common assessment task (CAT 2) were randomly selected and examined. The purpose of the study was to investigate in repair and non repair sequences types of assessor accommodation, how the assessors modified their utterances, the kinds of trouble perceived by assessors, what triggered assessor accommodation and whether the accommodations facilitated student response and participation. The accommodation types used by assessors were slow down, repetition, rephrase, elaboration, either/ or, tag and yes/no questions and supplying correct language forms. It was found that in repair sequences triggers of assessor accommodation were student lack of understanding, non response and misunderstanding. In these cases the trouble source was in the assessor talk and so the assessor perceived that some modification was needed to his/her utterance in order to remedy the trouble. At times assessors also repaired student talk triggered by problems with the students' use of the Italian language system. In non repair sequences triggers were in the talk of the student such as minimal answers, pauses and perturbation, and also previous accommodations triggering the assessors to continue accommodating. Many of the accommodation types were successful in ensuring students' responding appropriately with rephrasing and elaboration being the most successful since the trouble source was clarified. The accommodation type of repetition was found not to facilitate the subsequent student talk since the trouble source was not identified. This study has implications for assessor training since it highlights which strategies are most successful for ensuring student understanding, participation and appropriate responses as well as demonstrating why and in which environments assessors accommodate.