Measurement of glucose turnover under non-steady-state conditions has proven problematic. When the mass of the glucose pool is not changing (i.e., glucose concentrations are constant) non-steady-state error can be minimized if all glucose entering the circulation has the same specific activity as plasma [radioactive infused glucose (hot-GINF) method]. Alternatively, a second tracer can be used to measure the effective volume of glucose [variable-pV method of Issekutz (T. Issekutz, R. Issekutz, and D. Elahi. (Can. J. Physiol. 52:215-224, 1974)]. To determine whether these techniques provide concordant assessments of insulin action under non-steady-state conditions, glucose turnover was measured in six subjects. After initiation of insulin (0.6 mU.kg-1 x min-1), both methods indicated similar rates of suppression of hepatic glucose release, which was complete by approximately 100-120 min. In contrast, the traditional fixed-pV method of Steele (R. Steele, J. Wall, R. DeBodo, and N. Altszuler. Am. J. Physiol. 187:15-24 1956) underestimated turnover (P < 0.01) resulting in apparent complete suppression of glucose release within approximately 40 min (P < 0.01 vs. other methods). The hot-GINF and variable-pV methods also yielded similar estimates of turnover after discontinuation of insulin. Both indicated that resumption of hepatic glucose release was slower (P < 0.01) and fall of glucose uptake faster (P < 0.01) than suggested by the fixed-pV method. Thus both the hot-GINF and variable-pV methods avoid non-steady-state error introduced by the fixed-pV method and provide concordant assessments of the rate of onset and offset of insulin action.