Social identity theory predicts an interdependence between the relative evaluation of a person's in‐group, his or her social identity, and the degree of his or her in‐group‐out‐group differentiation. Three hypotheses were tested in an experiment that extended the original minimal group paradigm to one further out‐group as a second comparison group. The results show that an experimental devaluation of the in‐group in comparison to a first out‐group: (a) led to a more negative evaluation of those areas of the self which are substantially connected to the dimension on which the intergroup comparison took place (hypothesis 1); (b) did not result in a higher mean devaluation score for the second out‐group, which is inconsistent with hypothesis 2a; and (c) made the comparison dimension less important to the in‐group, thereby supporting hypothesis 2b. In an internal analysis, self‐evaluation in those aspects which relate to the intergroup comparison proved to be negatively correlated with the devaluation of the second out‐group. Hypothesis 3 proposed that self‐esteem would increase following devaluation of the second out‐group. The data pattern with respect to this prediction sho