Investigating the role of preferences in causing persistent patterns of racial residential segregation in the United States has a long history. In this paper, we bring a new perspective--and new data from the 2004 Detroit Area Study--to the question of how best to characterize black and white preferences toward living in neighborhoods with people of different races. White and black residents of the Detroit metropolitan area (n = 734) were asked in an area probability sample survey about their evaluations of 33 actual communities throughout their metro area. These evaluations are used as an indirect measure of racial residential preferences by viewing how race--both of the respondent and of the community--shapes them. We find modest racial agreement about which communities would be "seriously considered" and "never considered" as a place to live, but by and large perceptions of the metropolis are racialized. Whites are influenced by the percentage white in a community (net of the community's social class characteristics) and very unlikely to consider communities where they are anything but the strong majority. African Americans are also influenced by race, but in different ways and less fundamentally: 1.) Communities with high percentages of African Americans are among those most likely to be "seriously considered," but so are communities with just a handful of African Americans; 2.) African Americans are less likely to "never consider" all communities, and more likely than whites to consider both communities where they are in the majority and in the minority; 3.) African Americans are unaffected by a community's percent white net of community social class characteristics. We place these results in the context of the debate about racial residential preferences, arguing for the importance of grounding our understanding--and measures--of racial residential preferences in the context of real urban landscapes. (Contains 4 figures, 3 tables and 19 notes.)