Aortic stiffness, assessed by carotid-to-femoral pulse wave velocity (PWV), often fails to predict cardiovascular (CV) risk and mortality in the very elderly. This may be due to the non-linear association between PWV and compliance or to blood pressure decrease in the frailest subjects. Total arterial compliance (C-T) is the most relevant arterial property regarding CV function, compared to local or regional arterial stiffness. A new method for C-T estimation, based on PWV, was recently proposed. We aimed to investigate the value of C-T to predict all-cause mortality at the elderly. PWV was estimated in 279 elderly subjects (85.5 +/- 7.0 years) who were followed up for a mean period of 12.8 +/- 6.3 months. C-T was estimated by the formula C-T=kxPWV(-2); coefficient k is body-size dependent based on previous in silico simulations. Herein, k was adjusted for body mass index (BMI) with a 10 % change in BMI corresponding to almost 11 % change in k. For a reference BMI=26.2 kg/m(2), k=37. Survivors (n=185) and non-survivors (n=94) had similar PWV (14.2 +/- 3.6 versus 14.9 +/- 3.8 m/s, respectively; p=0.139). In contrast, non-survivors had significantly lower C-T than survivors (0.198 +/- 0.128 versus 0.221 +/- 0.1 mL/mmHg; p=0.018). C-T was a significant predictor of mortality (p=0.022, odds ratio=0.326), while PWV was not (p=0.202), even after adjustment for gender, mean pressure and heart rate. Age was an independent determinant of C-T (p=0.016), but not of PWV. C-T, estimated by a novel method, can predict all-cause mortality in the elderly. C-T may be more sensitive arterial biomarker than PWV regarding CV risk assessment.