According to sociological conventional wisdom, contact of the right kind in the right setting reduces stereotyping. Some research supports this hypothesis, some does not, and most can be challenged because the hypothesis conditions were not clearly met. This study attempted to meet the necessary conditions by using an unusually suitable setting, as well as refined measures of contact and of age, gender, and ethnic stereotyping. Interviews were conducted of 542 bridge players at the Ottawa-Hull branch of the American Contract Bridge League. Measures of contact were well indicated by the number of fellow players with whom a player was acquainted. Dimensions of age, gender, and ethnic stereotyping consisted of a set of semantic differentials. Respondents rated: (1) women as bridge players, men as bridge players, and women in general; (2) older people as bridge players, younger people as bridge players, and older people in general; (3) French Canadians as bridge players and as people in general. All measures were self-reports except for one measure of skill. The results failed to support the hypothesis that contact reduces stereotyping. Neither the frequency and duration of activity nor the proportion of group members in weak and in strong ties reduced stereotyped perceptions. The findings suggest that involvement in the core of the subculture actually increases stereotyping in that core players were more willing than periphery players to give ratings for little-known groups and to give extreme ratings for well-known groups. Greater subcultural involvement, experience, and communication only seemed to give greater confidence in applying general stereotypes to the specific setting. (NB)