1. Vavere, Andrea L. MS
  2. Foster, Carl PhD
  3. Brice, Glen PhD
  4. Martinez, Raymond PhD
  5. Porcari, John P. PhD

Article Content


Maximal HR (HRmax) is critical for exercise prescription. HRmax is usually predicted using age based equations, 220-age being the most common. The accuracy of 220-age is suspect, with SEE ~12 bpm. Even contemporary age based equations (averaging 210-0.70*age) have SEE of 6-13. This inaccuracy can lead to significant errors in exercise prescription.



This study was designed to test strategies for predicting HRmax based on the HR response during submaximal exercise.



Healthy volunteers (age 22-53) (n = 28) performed maximal treadmill exercise to define HRmax. They also performed submaximal exercise tests based on Edwards' Heart Zones approach. These submaximal tests were: 5-min Walk Test, 1-min Chair Test, 3-min Step Test, Talk Test, 2-min 10 Beat Test, 2 by 4-min Test, and Easy-Moderate-Hard Test. Each test had unique instructions and included a self-assessed fitness level to predict HRmax.



There were significant (P <.05) differences between measured HRmax (189 +/- 12) and predicted HRmax from: 2 x 4-min test (214 +/- 13), Talk Test (199 +/- 11), Step Test (202 +/- 20), Walk Test (176 +/- 22) and 2-min 10 Beat Test (214 +/- 8). Linear regression demonstrated that predicted HRmax from all submaximal exercise tests was poorly related to HRmax (R2 = 0.04-0.56). If self assessed fitness level was ignored, the highest HR during the submaximal tests was better in predicting HRmax (R2 = 0.01-0.66). The best predictors of HRmax were HR the last time speech was comfortable (R2 = 0.26; SEE = 12), HR at RPE = 5 (hard) (R2 = 0.34; SEE = 9), and HR at the time speech became impossible (R2 = 0.66; SEE = 7). Attempts to use the regression of RPE and HR during the incremental exercise test demonstrated that HR @ RPE of 5 (hard) was best at predicting HRmax (R2 = 0.21; SEE = 10).



HRmax is difficult to predict based on results of submaximal exercise tests. The gain in predictive accuracy over an optimized age prediction is not likely to be of practical significance.