Aquaporin-2 (AQP2) missense mutants in recessive nephrogenic diabetes insipidus (NDI) are all retained in the endoplasmic reticulum (ER), but some could function as water channels. No conclusions could be drawn about the water permeability (Pf) of others, because there was no method for quantifying AQP2 expression in the plasma membrane. We recently developed such a method, which has allowed us to study the functionality of these AQP2 mutants. Immunoblot analysis of membranes of injected oocytes revealed that all mutants (AQP2-G64R, AQP2-N68S, AQP2 T126M, AQP2-A147T, AQP2-R187C, AQP2-S216P) are expressed as unglycosylated and high-mannose glycosylated AQP2. The level of the high-mannose form of AQP2-A147T in the plasma membranes was low, indicating that this mutation has a less severe effect on proper folding. Analysis of Pf values and plasma membrane expression levels reveals that AQP2-N68S, AQP2-R187C and AQP2-S216P are non-functional, AQP2-A147T is as functional as wt-AQP2, while AQP2-T126M and AQP2-G64R retain 20% of the permeability of wt-AQP2. Since G64 is highly conserved between AQPs and expected to form essential interactions with other amino acids within AQP1, the residual functionality of AQP2-G64R is surprising. Our data furthermore indicate that an eventual therapy with chemical chaperones that restores the routing of AQP2 mutants to the apical membrane of collecting ducts cells might relieve NDI in patients encoding AQP2-A147T, and to a lesser extent AQP2-T126M and AQP2-G64R, but not in patients encoding AQP2-N68S, AQP2-R187C or AQP2-S216P.