The accuracy of fractional excretion of sodium (FENa) for the diagnosis of transient acute kidney injury (AKI) caused by decreased kidney perfusion is reported to be low in patients administered diuretics. This is a prospective study of diagnostic accuracy comparing the performance of fractional excretion of urea (FEur) with that of FENa to distinguish between transient and persistent AKI. 99 patients hospitalized at a tertiary-care center who developed AKI (>or=30% increase in serum creatinine level from baseline within 1 week). FEur and FENa were calculated for each patient. Patients were classified as having transient or persistent AKI according to the clinical context and whether serum creatinine level returned to baseline within 7 days. Each group also was subdivided according to exposure to diuretics. FEur of 35% or less and FENa of 1% or less were used to define transient AKI. Sensitivity, specificity, and receiver operating characteristic curves were generated for each index test. Sensitivity and specificity of FEur were 48% and 75% in patients not administered diuretics and 79% and 33% in patients administered diuretics. Sensitivity and specificity of FENa were 78% and 75% in patients not administered diuretics and 58% and 81% in those administered diuretics. Receiver operating characteristic curves did not identify a better diagnostic cutoff value for FEur or FENa. Small sample size, variable exposure to diuretics, and a high proportion of preexisting chronic kidney disease. In patients without diuretic use, FENa is better able to distinguish transient from persistent AKI. In patients administered diuretics, this distinction cannot be made accurately by means of FENa. FEur cannot be used as an alternative tool because it lacks specificity.