Background The Borg scale is most commonly used to measure dyspnea in China. However, many patients that find it is difficult to distinguish the labeled numbers corresponding to different dyspnea scores. We developed a new method to rate dyspnea, which we call the count scale (CS). It includes the count scale number (CSN) and count scale time (CST). The aims of the present study were to determine the reproducibility and sensitivity of the CS during exercise in patients with chronic obstructive pulmonary disease (COPD). Methods Fourteen male patients with COPD (aged 58.00+/-7.72 years) participated in this study. A progressive incremental exercise and a 6-minute constant work exercise test were performed every 2 to 3 days for a total of 3 times. The CS results were evaluated at rest and at 30% and 70% of maximal workload (W-max) and W-max. The Borg scales were obtained during exercise. Results No significant differences occurred across the three trials during exercise for the CS and Borg scores. The CSN and CST were more varied at W-max (coefficient of variation (CV)=(22.28+/-16.96)% for CSN, CV=(23.08+/-19.11)% for CST) compared to 30% of W-max (CV=(11.92+/-8.78)% for CSN, CV=(11.16+/-9.96)% for CST) and 70% of W-max (CV=(9.08+/-7.09)% for CSN, CV=(12.19+/-12.32)% for CST). Dyspnea ratings with either CSN or CST tended to decrease at the higher workload compared to the lower workload. CSN and CST scores were highly correlated (r=0.861, P <0.001). CSN was negatively correlated with Borg scores (r=-0.363, P=0.001). Similar results were obtained for the relationship between CST and Borg scores (r=-0.345, P=0.003). Conclusion We concluded that the CS is simple and reproducible when measuring dyspnea during exercise in patients with COPD.