Consensus refers to the extent to which judges agree in their ratings of a common target. This is a hotly debated area, both empirically and theoretically. The purpose of this study was to examine how consensus would wary as a function of a) how observable traits are, and b) acquaintance. From a critical realist viewpoint, this thesis argues that consensus will assume high values when well operationalized personality inventories are used within a fully crossed design. The data was collected from a leadership program at the Norwegian school of management; this study is a direct result of personality inventories collected there from 1993-2005. The percentage of total variance accounted for by target variance was used as the measure of consensus. Personality judgements from 126 groups, each including 8-11 persons, was analysed. In this sample, the personality inventory showed high levels of structural validity and construct validity. At zero acquaintance, substantial consensus was established for observable traits relating to extraversion. With higher levels of acquaintance, consensus increased substantially for all big five factors, extraversion rising to the level of .65. Consensus seemed to reach a maximum level, seemingly dependent of the quality of the personality relevant information available, and the actuality of different aspects of personality. The findings are discussed within an elaborate evolutionary framework.