This paper focuses on facework used by the invited guests in a broadcasted debate. The purpose of this analysis is to investigate how the invited guests use facework and how the guest relate to their functional role within the debate. The aim is also to investigate how the phases of the debate regulate the facework used by the guests. The analysis is based on transcrition and videomaterial from one of Sweden’s most viewed debates, Opinion Live. Conversation analysis (CA) is part of the theoretical framework used to analyse support, interruption and conversation turns. To analyze facework this paper relies on Goffman’s theory of face and Muntigl & Turnbull’s four types of disagreement acts in facework and arguing. Brown & Levinson’s politeness theory has also been used to investigate facework and strategies. In addition, Svensson’s definition of the debate’s specific roles within the format has also been used. The results show that facework has two purposes. The first is that it is used to defend the guest’s own face. The other purpose is to attack other guest’s faces which implies that their functional roles in the debate are being challenged. The results also points towards that the structure of the conversation in the debate itself influence the types of facework used by the guests.