Corruption is considered a relevant problem both locally and at a global level. Journalism has a fundamental role in curbing corruption, and levels of corruption can be significantly influenced by how the news media play their watchdog role. However, the way in which the media represent corruption is a relatively under-explored area of corruption studies. As an arena for public debate, media are central for the public understanding of complex, contested issues. As active participants in the public debate, they have the power to exert influence over the political and social life of a country. Through a multi-layer research design, this thesis explores the role of the media in the social construction of corruption in New Zealand and Italy, two countries characterised by very different levels of corruption. According to Transparency International’s Corruption Perceptions Index, New Zealand is among the least corrupt countries in the world, while Italy is among the most corrupt countries in Europe. Through a framing analysis of the news media coverage of two corruption scandals, and of the coverage of the yearly Corruption Perception Index, this thesis explores how corruption is differently framed in the two countries, focusing in particular on the different use of conceptual metaphors, personifications, metonymies and narratives. Results show that Italy is dominated by a “systemic corruption” frame, characterised by disease, war and disaster metaphors, and by a tendency to dilute or externalise responsibilities. On the other hand, New Zealand is characterised by a “corruption as individual crime” frame focused on prevention, in which corruption is constructed as a responsibility of individuals, in opposition to an integrity embodied in society and institutions. The empirical research is completed by an analysis of a debate developed in the New Zealand media over the practices of lafo and koha (Polynesian gift-giving traditions) in the context of New Zealand politics. Results show that by entering the framing contest, news media can have a powerful effect in developing informal and formal rules to regulate contested issues and grey areas of corruption. This empirical research offers new insights into the social construction of corruption in different contexts, in particular by using a comparative perspective on most different cases, and shifting the attention towards the construction of corruption in a context characterised by high levels of integrity. Theoretically, the empirical research is informed by a social constructionist perspective on framing. In opposition to recent claims for limiting the field of framing to the study of cognitive effects of equivalenced-based media framing, this thesis argues for a radically different perspective that sees framing as an eminently social process. Drawing from Serge Moscovici’s theory of social representations, some conceptual tools to overcome unclear points and missing gaps in framing theory are introduced. Firstly, this thesis shifts the attention from the debated question of the location of frames, to a more fundamental issue of the nature of frames. It is argued that frames are products of social processes, rather than elements of media content or human cognition. Moreover, a categorisation of frames’ dynamism is suggested, separating the two mechanisms on internal and interactive dynamism. Exemplifying it with results from the framing analysis of the media coverage of corruption, anchoring and objectification (drawn from social representations theory) are introduced as explanatory mechanisms of how framing devices are linked to their frames. Finally, a re-definition of frames that integrates conceptual and operational concepts, and supports a social constructionist perspective on framing, is suggested.