Despite the popularity of the concept of "anaerobic threshold" (AT), the noninvasive detection criteria remain subjective, and invasive validations of AT have been based on lactate data of arterial, mixed venous, venous, and capillary blood samples without any concern for the possible lactate differences from these sources. Eight normal subjects underwent two exercise tests on a bicycle ergometer. The protocol consisted of 3 min of rest, 3 min of 0 work load, and a 20 W/min ramp (1 W/3 s) until exhaustion. Simultaneous arterial and venous blood samples were drawn during the second test. Noninvasive gas response data were measured using a computerized breath-by-breath stress test system. Threshold phenomenon of the lactate accumulation was not found. The arterial lactate levels increased continuously after the start of the exercise ramp. The rise in venous lactate lagged behind the rise of the arterial lactate by about 1.5 min, and therefore venous lactate was not considered suitable for AT detection. Four independent exercise physiologists determined AT from the gas response data. The reviewer variability (avg range 16%) of AT for a given subject was representative of AT values reported for untrained and trained individuals (40-70% maximum O2 consumption). We concluded that 1) AT is not detectable using invasive methods (arterial and venous lactates); and 2) the noninvasive gas response determination has such a large range of reviewer variability that it is unsuitable for clinical use.