An approach to translation is discussed which takes into account the sociolinguistic factors involved in a translated document designed for use with a linguistically, culturally and educationally heterogeneous population, such as the U.S. Spanish-heritage community. The translation project described resulted in the Spanish-language questionnaire for the Survey of Income and Education (SIE), which was conducted during the spring of 1976. As background, the language and cultural variation among Spanish-speaking groups is discussed. Because of this diversity, it was felt that the questionnaire should evidence sociolinguistic sensitivity. The project itself had two objectives: to make the Spanish both appropriate and acceptable to most of the Spanish-speaking populations in the United States, and to make the translation as useful as possible to the Bureau of the Census. The consensus translation process, by which these objectives were achieved, is described in detail. It involved: (1) the selection of a team of translators from different U.S. Hispanic communities who were trained linguists experienced in translating; (2) initial translation of the SIE documents into Spanish by one translator; (3) back-translation of the documents into English by all the translators; and (4) a series of meetings of the translation team to develop and refine the translation. Issues and problems concerning format and style are also described, and guidelines used in their resolution are given. The consensus translation process demonstrated in this project is felt to provide a potential model for all large-scale surveys involving Spanish-speaking groups. It is recommended that several needs discovered during the project (such as specialized glossaries and special training for bilingual interviewers) be explored to make the consensus translation process more effective. (AMH)