Investigates an old controversy in ethnic identification from the perspective of information-gathering strategies. It was hypothesized that people would request a lot of positive information before deciding that someone is a member of the ingroup. First, a questionnaire measuring the typical features of likeable and unlikeable targets issuing from two linguistic groups (Flemish and Walloon) revealed the existence of four distinguishable sets. These sets corresponded to the orthogonal combination of valence and group membership, i.e. they were organized in terms of two independent dimensions, an evaluative one and descriptive one. The dimensional complexity and evaluative extremity of the ‘positive ingroup’ and ‘negative outgroup’ sets were not different. Second, characteristics in each set served to create personality profiles presumably describing real targets. Subjects read these profiles, one feature at a time up to 10 features, and were asked to decide whether the target was a member of their group. They also learned that they could make their decision as soon as they felt confident. In line with Yzerbyt and Leyens' (1991) results, data indicate that subjects requested more information when the evidence was positive or consistent with their ingroup membership than when it was negative or inconsistent. These findings shed new light on earlier work concerning ethnic identification. In the context of the more general question of intergroup relations and their role in person perception, the present results may be interpreted in terms of an ingroup overexclusion effect rather than a vigilance effect or response bias. Thus is added a new effect to the well-known phenomena of ingroup favouritism and outgroup homogeneity.