Despite their longstanding interest in race, American sociologists have conducted little empirical research on sociodemographic patterns or longitudinal trends in "racial conceptualization" -that is, notions of what race is, how races differ, and the origins of race. This article outlines key empirical, methodological and theoretical considerations for a research agenda on racial conceptualization. Drawing on in-depth interviews with more than 50 college students, I describe the variety of race concepts among respondents, illustrate the importance of using multiple measures of conceptualization, and demonstrate the malleability of conceptualization, linking it to demographic context and thereby raising the question of its future evolution in the changing United States of the 21 st century.The color line, "problem of the twentieth century" as Du Bois (1986[1903]) famously put it, has long been a prominent concern of American sociologists (Calhoun 2007). The ways in which they have engaged the topic of race, however, reflect the preoccupations of their times. Early work on "race relations" (Park 1949) gave way to theories of "racism" in the civil-rights era, drawing new attention to institutional structures of racial oppression (Winant 2000). Large-scale surveys began to track attitudes -toward groups and policies -that might pose obstacles to achieving racial equality (Schuman, Steeh, Bobo and Krysan 1997). And in the wake of diversifying immigration inflows and rising intermarriage rates, scholars have revisited longstanding assumptions about racial identity and classification, launching new research on the categorization of mixed-race people and immigrant groups (Lee and Bean 2004). By the end of the 20 th century, American sociology had acquired a significant body of knowledge on race relations, attitudes, stratification and classification. I thank Elizabeth M. Armstrong, Troy Duster, Michèle Lamont, François Nielsen, Marta Tienda, Robert Wuthnow and an anonymous reviewer for thoughtful comments on
