Recent work indicates that trying not to think in stereotypical terms increases the accessibility of stereotypical information, which paradoxically results in more stereotypical judgments. The present study translated the colour-blindness ideology in general and stereotype suppression research in particular into an hypothesis testing setting. Participants who were asked to suppress their stereotypes when selecting a set of questions were indeed less guided by ambient stereotypes than control participants, thereby showing a reduction of the classical confirmation orientation in question preferences. Still, compared to control participants, suppressors also later reported more polarized impressions such that consistent targets were seen as more stereotypical and inconsistent ones as more counter-stereotypical. Moreover, group evaluations were more stereotypical for suppressors than for controls indicating that suppression had led to stronger activation of the stereotypical representation. Results are discussed in light of the prevailing belief regarding the benefits of political correctness and colour-blindness.