This article examines how 2000 US census statistics serve to reproduce Arab Americans as a social body with specific political possibilities. Its first goal is to show how the conventionally used census data acts as a source of statistical and, furthermore, social and political marginalization by rendering Arab Americans, who poorly fit into its racial classification, either invisible or lacking in internal heterogeneity. The result is absent viable political identity for Arab Americans, general ignorance of their experiences in the United States, and negative stereotyping. Because in the post-9/11 United States, Arab American advocacy groups have turned to statistics, among other means, to forge positive visibility, our second goal is to suggest innovative and practical ways to make a more informed use of the existing census statistics despite that their categorical and numerical inconsistencies relative to Arab American identities. Instead of a quantitative analysis, we thus offer a feminist inspired method of " mapping for difference" three census datasets-Arabic language spoken at home, place of birth in an Arab nation, and Arab ancestry, which involves applying a primarily qualitative analysis to these variables while contextualizing them by immigration history. Our result is a set of heterogeneous Arab American geographies, not intended for better quantitative description but serving to counter practices of marginalization by broadening public imagination and knowledge about this diverse community. Throughout, we suggest that while the designation Arab American maybe practically and politically useful, the recognition of the heterogeneity of Arab community along multiple dimensions of difference must be built into the methods of analysis. We demonstrate our major points with selected empirical maps from our larger research project on the census-based geographies of Arab Americans in the New York metropolitan area, one of the largest and most understudied communities in North America.