The future of endangered languages is discussed, drawing on experiences in development of an atlas of language. Issues in the debate over language maintenance are examined, including the emotion-laden concept of ethnicity and relative youth of the concepts of human rights and linguistic geography. The fact that the atlas in question indicates only indigenous language use is also discussed, and cartographic reasons are noted. Description of the process used to create the atlas' first edition exemplifies the difficulty of indicating languages with small or widely dispersed populations. The case of Canada is used for illustration of the occasional conflict between synchronic and diachronic perspectives in atlas development, including the French/English situation in Quebec, shrinking indigenous language groups, widely varying population densities and composition, difficulties in gathering accurate and timely information, and extralinguistic factors outside the researcher's expertise or prediction. The case of the Livonian language is offered as an example of an endangered language affected by international politics. Issues in language use prediction are then addressed, including difficulties in monitoring change and creating an adequate corpus for reference. Criteria for assessing language health are listed, and a brief continent-by-continent survey of languages is outlined. (MSE)