Our paper presents a survival analysis of the mortality follow-up to 2007 of the 1999/2000 Italian National Health Interview Survey sample. Kaplan-Maier curves illustrate survival patterns during the period of follow-up of people with different health conditions and educational levels at the moment of interview, while Cox regression models are estimated to quantify the strength of the association between poor health, low education and mortality. Both functional and subjective health are considered, as measured by presence/absence of limitations in Activities of Daily Living (ADLs) and Self-Rated Health (SRH) respectively. All analyses are performed on the subsample of individuals aged 50 years and over, excluding those who died from violent or external causes. Results indicate that: 1) the effects of severe limitations in ADLs and poor SRH on mortality change over time, being strongest immediately after the interview and then decreasing over the whole period of follow-up; 2) the association of limitations in ADLs and poor SRH with mortality is significantly stronger among people with medium-high rather than low education. The latter point supports the hypothesis that education plays a moderating role in the relationship between poor health and mortality.