OBJECTIVE To determine whether the lactate threshold of dogs could be determined by a visual method and to assess the extent of agreement and bias among treadmill velocities for the lactate threshold as determined by visual (LTv) and polynomial (LTp) methods, glucose threshold as determined by visual (GTv) and polynomial (GTp) methods, and heart rate deflection point (HRdp) as a method for estimating the aerobic capacity of dogs. ANIMALS 18 healthy adult Beagles. PROCEDURES Each dog underwent a standardized incremental treadmill exercise test once. The test ended when the dog began to show signs of fatigue. Plasma lactate and glucose concentrations and heart rate (HR) were plotted against exercise intensity (treadmill velocity) for the duration of the test, and the LTv, GTv, and HRdp were determined visually. The LTp and GTp were determined by means of a second-order polynomial function. One-way ANOVA, Pearson correlation, Bland-Altman analyses, and ordinary least products regression were used to assess the extent of agreement and bias among the various threshold velocities. RESULTS Mean velocity did not differ significantly among the thresholds evaluated. There was a strong positive correlation between the LTv velocity and the velocity for GTv (r = 0.91), LTp (r = 0.96), GTp (r = 0.94), and HRdp (r = 0.95). CONCLUSIONS AND CLINICAL RELEVANCE Results indicated that LTv could be determined for dogs undergoing intense exercise on a treadmill, and the treadmill velocity corresponding to the LTv was associated with the velocity for the other hallmarks of endurance. Thus, that method may be useful for prescription and evaluation of conditioning programs for dogs.