Between 1992 and 2040, the United States nonwhite elderly population is expected to grow from 3.3 to 14.1 million. In order to assess the implications of this increase on the mortality from neurodegenerative diseases in the United States, we used Census Bureau population estimates to formulate projections of the annual number of deaths from neurodegenerative diseases and from six comparison conditions (liver cirrhosis, colon cancer, lung cancer, cancer of the female breast, multiple sclerosis, and malignant melanoma), assuming that the United States disease-age-gender-specific death rates for 1985–1988 remain constant between 1990 and 2040. We find that neurodegenerative disease mortality increases by 281–524%, depending on the model of population growth used. For the ''middle'' population growth model, the increase in annual neurodegenerative disease mortality is 373%. The major component of this increase is the rise in deaths attributed to dementia. For the six comparison diseases, the increases in mortality range from 130 (multiple sclerosis) to 288% (colon cancer). Given the current level of underascertainment of neurodegenerative disease mortality, particularly among minorities, and the conservative nature of the Census Bureau estimates of future population, it is likely that these projections are underestimates. The implications of these data are discus