Introduction: Heart rate (HR) has been a simple and easy-to-use physiological parameter widely used to determine exercise intensity. The critical power fatigue limit model, known as the critical heart rate (CHR), can be extrapolated to HR. However, an estimate for a CHR mathematical model has not yet been extrapolated for upper limb exercise in the elderly. Objective: To compare the mathematical model previously used to estimate CHR with the heart rate values at the critical power (CP) during arm-ergometer exercises in elderly subjects. Methods: After an initial maximum-incremental exercise test on a cycle arm-ergometer, seven elderly people performed four high-intensity constant-load tests to the limit of tolerance (Tlim), to determine CP and critical heart rate (CHR). For each power output, the heart rate of the last five seconds (HRlim) and total time to exhaustion (in minutes) were obtained. The slope coefficients of the regression lines between HRlim and Tlim were defined as CHR, and between Wlim and Tlim as CP. A square-wave test was performed on a different day, in the power determined as equivalent to CP, and the heart rate at CP (CPHR) was assessed. Results: The HR-Tlim relationship was found to be hyperbolic in all subjects, who were able to sustain upper-limb exercise at CP for 20 min. CP attained 66.8±9.4% of peak work rate in the ramp test. The real average HR measured in the CP test was strikingly similar to the CHR calculated by the mathematical model of PC (137.6±16.9 versus 139.7±13.3bpm, respectively, p=0.53). There was strong correlation between the real and the estimated CHR. Conclusion: This study indicated that the maximal sustainable exercise intensity can be based on a physiological variable such as HR, and the CHR test can define exercise endurance, which can be useful in performance assessment and training prescription. Level of evidence II; Diagnostic studies – Investigating a diagnostic test.