Urea production rate (Ra) is commonly measured using a primed continuous tracer urea infusion, but the accuracy of this method has not been clearly established in humans. We used intravenous infusions of unlabeled urea to assess the accuracy of this technique in normal, postabsorptive men under the following four different conditions: 1) tracer [13C]urea was infused under basal conditions for 12 h (control); 2) tracer [13C]urea was infused for 12 h, and unlabeled urea was infused from hours 4 to 12 at a rate twofold greater than the endogenous Ra ("step" infusion); 3) tracer [13C]urea was infused for 12 h, and unlabeled urea was infused from hours 4 to 8 ("pulse" infusion); and 4) tracer [13C]urea was infused for 9 h, and unlabeled alanine was infused at a rate of 120 mg.kg-1.h-1 (1.35 mmol.kg-1.h-1) from hours 4 to 9. Urea Ra was calculated using the isotopic steady-state equation (tracer infusion rate/tracer-to-tracee ratio), Steele's non-steady-state equation, and urinary urea excretion corrected for changes in total body urea. For each subject, endogenous urea Ra was measured at hour 4 of the basal condition, and the sum of this rate plus exogenous urea input was considered as "true urea input". Under control conditions, urea Ra at hour 4 was similar to that measured at hour 12. After 8-h step and 4-h pulse unlabeled urea infusions, Steele's non-steady-state equation underestimated true urea input by 22% (step) and 33% (pulse), whereas the nonisotopic method underestimated true urea input by 28% (step) and 10% (pulse). Similar conclusions were derived from the alanine infusion. These results indicate that, although Steele's non-steady-state equation and the nontracer method more accurately predict total urea Ra than the steady-state equation, they nevertheless seriously underestimate total urea Ra for as long as 8 h after a change in true urea Ra.