Aims Chronotropic incompetence and impaired heart rate (HR) recovery are related to mortality. Guidelines lack specific reference values for HR recovery. We defined normal values and studied blunted HR response and recovery and mortality risk. Methods and results We included 9917 subjects (45% females) aged 18-85 years who performed a cycle exercise test. We defined normal values for peak HR, HR reserve, and HR recovery at 1 and 2 min (HRR1 and HRR2) based on individuals apparently healthy (N = 2242). Associations between blunted HR indices (<5th percentile) and mortality over a median follow-up of 8.6 years were analysed using Cox regression and competing risk analysis. All HR indices were age-dependent and independent predictors of all-cause and cardiovascular (CV) mortality. The 5th percentiles of HR reserve, HRR1, and HRR2 correlated weakly with existing reference values. Heart rate recovery variables were the strongest predictors of all-cause mortality in both the overall population [HRR1, hazard ratio 1.70 (95% confidence interval, 1.49-1.94), and HRR2, 1.57 (1.37-1.79)] and in subjects with normal exercise capacity [HRR1, 1.96 (1.61-2.39), and HRR2, 1.76 (1.46-2.12)]. Combining HR indices appeared to increase the risk of all-cause [HRR1 and HRR2, 1.96 (1.68-2.29), and peak HR and HRR1, 1.87 (1.56-2.23)] and CV mortality, although no specific combination was superior for predicting CV mortality. Conclusion All HR indices were age-dependent and associated with all-cause and CV mortality. Blunted HR recovery variables were the strongest predictors of all-cause mortality, even in subjects with normal exercise capacity. Combined blunted HR indices appeared to add prognostic value.