The present study examined several theories regarding self-identification and racial preference among White and Native children. Construct accessibility theory predicts that cues, such as experimenter's race, prime race and race-related constructs, making them more accessible for processing race-related information. Both the escape and light colour bias hypotheses predict (for different reasons) that minority group children should identify with and show a preference for White children. To test these ideas White and Native children were asked by a White or Indian experimenter to answer questions about their racial identity and preferences. Subjects responded by pointing to a picture of a White boy, a White girl, a Native boy, or a Native girl. Consistent with predictions from accessibility theory, subjects of both races made more accurate self-identifications when tested by an Indian experimenter, this effect being greater for Native children. Subject's racial preferences, on the other hand, showed a strong White bias, results consistent with both the escape and the light colour bias hypotheses. The different pattern of responses to preference and self-identity questions was discussed in term of Zajonc's (1980) distinction between cognitive and affective processes. Applications of these findings to educational and cultural issues confronting Native people are presented.Nearly fifty years ago Clark and Clark (1939) found that Black children typically chose a White rather than a Black doll to play with, called White dolls nice, and viewed Black dolls as bad. From these and other findings Clark and Clark inferred that American Blacks reject their race. Subsequent studies have supported these findings (but not necessarily their interpretation), and have also indicated that they are not limited to Blacks; racial identification and preference among other visible minorities such as Native Indians, Chinese Americans, Chicanos, and New Zealand Maoris also demonstrates a consistent White bias (see Aboud & Skerry, 1984). For example, both George and Hoppe (1979) andHunsberger (1978) have shown that Canadian Native children, when presented with pictures or dolls of White and Native children, were more likely to misidentify which doll looked like them, to prefer the White doll as a playmate, and were less accurate than White age peers at indicating which picture or doll looked like a Native child. Other studies Our thanks are extended to Candis Caryk, Donna Gamble, Sonia Pressman, Lillian Bighetty, and Hilda Bighetty for serving as our experimenters. Special thanks are extended to the teachers and principals of the Brandon School System and to Peter Giesbrecht for his programming skills. We are grateful to Francis Aboud and Cliff Anderson for their comments on an earlier draft of the manuscript.