The present study assessed the relationship between coffee consumption and mortality in a home-dwelling elderly population. A population-based cohort of 817 men and women born in 1920 or earlier and living in northern Finland provided complete data on daily coffee consumption and other variables at the baseline examination in 1991–1992. Deaths were monitored through to the end of 2005 by national death certificates, resulting in 6960 person-years of follow-up. Hazard rate ratios for mortality by daily coffee intake were estimated by Poisson regression models adjusted for some known predictors of mortality. During 14·5 years of follow-up, 623 deaths occurred. The total mortality rate was inversely related to the number of cups (average volume, 125 ml) of coffee consumed daily. After adjustment for age, sub-period of follow-up, sex, marital status, basic educational level, previous occupational group, current smoking, BMI, history of myocardial infarction, self-rated health and presence of diabetes, cognitive impairment or physical disability, the estimated relative risk reduction of total mortality per an increment of one more cup of coffee per d reported at baseline was 4 (95 % CI 0, 8) %. The observed associations between coffee consumption and mortality from CVD, cancer, and other or unknown causes, respectively, were qualitatively similar to that of total mortality but the estimates were less precise. The effect of coffee consumption at baseline appeared to attenuate after 10 years since the start of follow-up. The present study provides evidence for daily (caffeine-containing) coffee intake being inversely associated with mortality in the elderly.