This study used a signal detection paradigm to explore the Marley hypothesis--that group differences in perception of racism reflect dominant-group denial of and ignorance about the extent of past racism. White American students from a midwestern university and Black American students from two historically Black universities completed surveys about their historical knowledge and perception of racism. Relative to Black participants, White participants perceived less racism in both isolated incidents and systemic manifestations of racism. They also performed worse on a measure of historical knowledge (i.e., they did not discriminate historical fact from fiction), and this group difference in historical knowledge mediated the differences in perception of racism. Racial identity relevance moderated group differences in perception of systemic manifestations of racism (but not isolated incidents), such that group differences were stronger among participants who scored higher on a measure of racial identity relevance. The results help illuminate the importance of epistemologies of ignorance: cultural-psychological tools that afford denial of and inaction about injustice.