When people are interested in how common one or more of their attributes is in a reference population, they must often generate their own comparison information based on a limited sample of acquaintances and experiences. Subjects in the present research were asked to describe themselves in terms of a variety of attributes, and were also asked to estimate the percentage of other college students who would indicate possession of each attribute. For each attribute, subjects were assigned a majority or minority status, depending on whether the majority of the population did or did not share their attribute. The principle findings were (1) the majority subjects generated more accurate comparison information than did the minority; (2) the majority subjects were better than chance and better than minority subjects in distinguishing between attributes for which there was high versus moderate consensus; and (3) the minority subjects tended to overestimate the consensus for their attributes, while the majority subjects tended to err in the direction of underestimation of their Consensus. The discussion focused on possible causes of these tendencies, and on research implications involving attributional biases and intergroup conflict.