This study tested the ventilation (VE) behavior during upper-body incremental exercise by mathematical models that calculate 1 or 2 thresholds and compared the thresholds identified by mathematical models with V-slope, ventilatory equivalent for oxygen uptake (VE/V(O2)), and ventilatory equivalent for carbon dioxide uptake (VE/V(CO2)). Fourteen rock climbers underwent an upper-body incremental test on a cycle ergometer with increases of approximately 20 W · min(-1) until exhaustion at a cranking frequency of approximately 90 rpm. The VE data were smoothed to 10-second averages for VE time plotting. The bisegmental and the 3-segmental linear regression models were calculated from 1 or 2 intercepts that best shared the VE curve in 2 or 3 linear segments. The ventilatory threshold(s) was determined mathematically by the intercept(s) obtained by bisegmental and 3-segmental models, by V-slope model, or visually by VE/V(O2) and VE/V(CO2). There was no difference between bisegmental (mean square error [MSE] = 35.3 ± 32.7 l · min(-1)) and 3-segmental (MSE = 44.9 ± 47.8 l · min(-1)) models in fitted data. There was no difference between ventilatory threshold identified by the bisegmental (28.2 ± 6.8 ml · kg(-1) · min(-1)) and second ventilatory threshold identified by the 3-segmental (30.0 ± 5.1 ml · kg(-1) · min(-1)), VE/V(O2) (28.8 ± 5.5 ml · kg(-1) · min(-1)), or V-slope (28.5 ± 5.6 ml · kg(-1) . min(-1)). However, the first ventilatory threshold identified by 3-segmental (23.1 ± 4.9 ml · kg(-1) · min(-1)) or by VE/V(O)2 (24.9 ± 4.4 ml · kg(-1) · min(-1)) was different from these 4. The VE behavior during upper-body exercise tends to show only 1 ventilatory threshold. These findings have practical implications because this point is frequently used for aerobic training prescription in healthy subjects, athletes, and in elderly or diseased populations. The ventilatory threshold identified by VE curve should be used for aerobic training prescription in healthy subjects and athletes.