The prognostic value of heart rate variability (HRV) in patients with cardiac conditions has been investigated for many years. However, the HRV is superior to annual health examinations for predicting the longevity of very elderly residents of long-term care facilities is unknown. Annual health examinations and subsequently ambulatory Holter ECG recording were performed in 2008 for 71 very elderly subjects, who were then followed up for 3 to 48 months. The patients were divided into 2 groups on the basis of whether they were alive (86 ± 14 years, n=37) or deceased (90 ± 16 years, n=34) at end of follow-up. To assess cardiac autonomic function, HRV was obtained with the MemCalc/Chiram software program after Holter ECG. Age, sex, body-mass index, plasma levels of C-reactive protein and albumin, and the low-frequency/high-frequency ratio did not differ between the 2 groups. However, the standard deviation of all NN intervals (SDNN) and the coefficient of variation of RR intervals (CVRR) were higher in living subjects than in deceased subjects (SDNN: 73.2 ± 13.5 milliseconds vs. 53.2 ± 9.8 milliseconds, CVRR: 9.3% ± 1.7% vs. 7.6% ± 1.3%, p<0.05). The relative risks with an SDNN <65 milliseconds was 1.85 (p<0.05) and that with a CVRR <8% was 1.84 (p<0.05). Kaplan Meier analysis showed that SDNN and CVRR were useful markers for the longevity of very elderly subjects. The present data suggest that annual health examination data does not predict longevity, but that HRV does. The modulation of parasympathetic tone in daily activities plays an important role in the longevity of very elderly residents of long-term care facilities.