This study explores the context-dependent nature of perceptions of group variability by examining how ingroup and outgroup ratings are affected by asking participants (N = 237) to rate these groups either on their own or together. In key respects, it replicates the design utilized by Haslam, Oakes, Turner, and McGarty (1995). However, several features of the present study's design were distinctive and intended to address methodological issues raised in the original. First, an alternative to the Katz-Braly procedure was adopted such that the measure of 'stereotypicality' referred to the applicability of both stereotypic and counter-stereotypic attributes. Second, a measure of 'dispersion' was also employed. The data show that ingroup stereo typicality was, as predicted, affected by a manipulation of context, while outgroup stereo typicality was not. No effects of context on the dispersion measures were obtained for either the ratings of the ingroup or the outgroup. The relationship between stereotypicality and dispersal measures of variability is discussed.