Over the past few decades, accountability has become a central concept in the study of the public sector. However, the concept has proven to be difficult to translate into various situations, considering the diverse ways in which accountability is constructed. Previous literature has provided insight of this diverse concept, as well as highlighting the importance of the accountability context. The present thesis focuses on exploring how accountability is constructed in public collaboration projects by turning to the process of people constructing relationships. Thus, the study goes beyond the concept itself to examine the different aspects of constructing accountability relationships within the context of collaborative dynamics.The empirical study focuses on three cases of collaboration projects initiated by county administrative boards and regions in Sweden. All three collaboration projects address societal and environmental problems, also known as "wicked problems". Despite having limited resources, these projects are driven by individuals who share their knowledge and enthusiasm. By examining the activities involved in 'collaboration work', the study illustrates how accountability is intertwined with context, expectations and engagement.By acknowledging the multifaceted process of constructing accountability, the analysis highlights how individuals create an 'accountability space', in which accountability becomes relevant and expressed. Furthermore, this study uses the concept of boundary action to explore how people make sense of accountability and visualise the different actions that contribute to its creation. The concept of boundary action assumes that accountability is created through time and space and is defined by the actions performed by individuals, and how these actions reflect social agreements within the context of the collaboration project. Contrary to previous ideas, these constructions of accountability demonstrate that questions such as 'who', 'what' and 'why' are rarely precise when accountability is constructed. Instead, this work contributes to highlight the asymmetric, dynamic and pluralistic dimensions of the social construction of accountability. This work reflects the multifaceted, diverse and variegated nature of accountability.