Within the framework of social identity theory (SIT) two experimental studies were conducted to further explore group members' selectivity in choosing dimensions for intergroup discrimination. Both studies were carried out in a laboratory setting using artificially created social categories. Importance of comparison dimension for the in‐group and that for the out‐group were manipulated independently of each other as within‐subject factors. In line with previous research, the first study tested the following hypotheses using 32 female respondents: (1) in‐group favouritism would be higher on dimensions highly important to the in‐group than on those unimportant to it; (2) in‐group favouritism would be mitigated on dimensions highly important on both in‐ and out‐group; (3) out‐group favouritism would occur on dimensions which are highly important to the out‐group but at the same time unimportant to the in‐group. All three hypotheses were confirmed. The second study added relative in‐group size (minority vs. majority position) as a between‐subjects factor to the experimental design. Respondents were 46 female students. It was predicted that minority members would discriminate in favour of their own group particularly on dimensions important to the in‐group, whereas majority members would favour the in‐group only on dimensions unimportant to the out‐group. The results partially supported our predictions. In addition, more subtle pathways to out‐g