The article discusses the history and ideology of the censuses of the population in the United States, a feature of which compared to, for example, the censuses held in European countries, is the use of two categories - race and ethnicity. At the same time, in determining official racial identities, the role of language is usually not taken into account. Historically, race included physical, hereditary, and invariable differences, and ethnicity was seen as a cultural difference, passed on from generation to generation, but could change over time. Therefore, racial discourse presents some groups as different and dangerous, and ethnic discourse depicts them as safe and even colorful. The main racial difference is based on the opposition of groups consisting of whites and groups consisting of non-whites, and racial differences are also defined within groups consisting of non-whites. Thus, censuses are a statistical reflection of a two-term socio-political structure, in which the opposition “white” to “non-white” was viewed as unchanged during the transition from one generation to another and having more significance than racial differences between whites, which disappear after the second generation. In recent decades, special attention has been paid to the ethnic origin of Hispanic Americans, which include Hispanics, Latinos and Spaniards. They, in turn, are counted in the following categories: Mexicans, Americans of Mexican Origin, Chicanos, Puerto Ricans, Cubans and others - Argentines, Colombians, etc.), but they do not constitute a separate race. The modern formulation of the question of languages in censuses is characterized by the fact that in it language behavior is put over hereditary ethno-racial or cultural identity.