This study examined the association between mortality and nonresponse to questions about health status (both refusals and "don't know" responses) using a national sample of persons aged 70 and over. Data were drawn from the 1984-1990 Longitudinal Study of Aging. Three time points of vital status were used as the outcome indicators (1984-1986, 1984-1988, 1984-1990). Five self-assessment questions were examined; three of the five questions had bivariate odds ratios that indicated significant associations between a nonresponse and all three mortality indexes. Results of the study suggest that nonresponses by older persons can convey meaningful information. Research on self-assessments of health in later life should not routinely exclude nonresponses as missing data, even if they are an infrequent response.