Knowledge of age profiles for functional impairment risk, and the corresponding profiles for disability-free or healthy life expectancy, as well as their evolution, is important for planning health and long-term care resources in economies with an aging population. In this paper we estimate such profiles for the years 1985, 1996 and 2007 using a longitudinal population health study in Norway. By paying particular attention to the issue of sample attrition, we find that when correcting for such selection issue, the increase between 1985 and 2007 in expected years (at age 50) with functional impairment is 3.1 (1.8) years above the corresponding figures obtained for women (men) when relying on raw sample prevalence rates. Most studies on healthy life expectancy rely on self-reported health surveys to measure the prevalence of function impairment, but few consider the implications of sample selection. Our results highlight the need to pay attention to such data imperfections.