We simulate a job application/hiring market in the lab to examine racial discrimination. We find little evidence of ability differences based on race but we find taste-based racism between groups and statistical racism within groups. When candidates are given the opportunity to lie about their abilities, all groups discriminate against Blacks, suggesting statistical discrimination. But Whites continue to discriminate against Blacks when actual abilities of the candidate are known, suggesting taste-based discrimination. In contrast to the bulk of studies that attempt to establish racism in general as either a taste-based or statistical, our design allows us to show that the type of discrimination can depend on the personal characteristics of the discriminating individual along with the contextual information available.