The ventilatory threshold (VT ; determined by gas exchange parameters) has been applied for many purposes regarding to physical working capacity in many studies. VT, however, contains the subjective variation, because it is determined from some criteria of gas exchange parameters visually. Recently, there are some reports that lactate threshold (LT)and VT do not always occur simultaneously. In this study, we employed the segmented regression analysis to estimate LT and VT objectively, and re-examine the relationships of LT computed, VT computed, and VT deterTnined visually. Thirteen young male subjects performed an incremental-load ergometer work in which the initial work rate was the 4 min unloaded cycling and thereafter the work rate was increased 150 kgm every 2 min until exhaustion. Gas exchange parameters were measured by Douglas bag method continuously, and blood samples for LA analysis were obtained last 30 sec each work rate from warmed ear lobe. It is supposed that LA and VE changes against Vo2 during progressive work is fitted by three segment model, so we employed the segmented regression analysis to all data and data below 85% Vo2max of LA and VE vs. Vo2, and estimated LT(all), LT(85%), VT(all), and VT(85%), respectively. VT(visual) was a mean value of four observers who visually inspected the graph of VE, VE/Vo2, and VE/Vco2 vs. Vo2 according to two criteria described below, 1) non-linear increase in VE, 2) increase in VE/Vo2 Without corresponding increase in VE/Vco2. The results of this study were 1) LT(85%) is always consistent with LT(all) and have a standard deviation of about 0.1L/min-Vo2, 2) all variables that computed or inspected by gas exchange informations and LT(85%) do not always occur and these results suggest limitations and invalidity of using gas exchange parameters to estimate LT.