Clinical studies have shown that aquaporin-2 (AQP2), the vasopressin-regulated water channel, is excreted in the urine, and that the excretion increases in response to vasopressin. However, the cellular mechanisms involved in AQP2 excretion are unknown, and it is unknown whether the excretion correlates with AQP2 levels in kidney or levels in the apical plasma membrane. The present study was undertaken to clarify these issues. Immunoblotting of rat urine samples revealed significant excretion of AQP2, whereas AQP3, being a basolateral aquaporin in the same cells, was undetectable. Thus, there was a nonproportional excretion of AQP2 and AQP3 (compared with kidney levels), indicating that AQP2 is excreted predominantly via a selective apical pathway and not by whole cell shedding. Urinary AQP2 was associated with small vesicles, membrane fragments, and multivesicular bodies as determined by immunoelectron microscopy and negative staining techniques. In rats with normal water supply, daily urinary excretion of AQP2 was 3.9 ± 0.9% (n = 6) of total kidney expression. Treatment with desmopressin acetate subcutaneously caused a fourfold increase in urinary excretion of AQP2 during 8 h. Forty-eight hours of thirsting, known to increase endogenous vasopressin secretion, resulted in a three-fold increase in kidney AQP2 levels but urinary excretion increased ninefold to 15 ± 3% (n = 6) of AQP2 in kidney of thirsted rats. Moreover, rats that were thirsted for 48 h and subsequently allowed free access to water for 24 h produced a decrease in urinary AQP2 excretion to 38 ± 15% (n = 6) of that during thirsting. In Brattleboro rats or lithium-treated normal rats completely lacking vasopressin action, and hence having extremely low levels of AQP2 in the apical plasma membrane, AQP2 was undetectable in urine. Thus, conditions with known altered vasopressin levels and altered levels of AQP2 in the apical plasma membrane were associated with corresponding major changes in AQP2 urine excretion. In contrast, in such conditions, kidney AQP2 levels and urinary AQP2 excretion did not show a proportional relationship.