Nephrogenic diabetes insipidus (NDI) is characterized by resistance of the kidney to the action of arginine-vasopressin (AVP); it may be due to genetic or acquired causes. Recent advances in molecular genetics have allowed the identification of the genes involved in congenital NDI. While inactivating mutations of the vasopressin V₂ receptor are responsible for X-linked NDI, autosomal recessive NDI is caused by inactivating mutations of the vasopressin-regulated water channel aquaporin-2 (AQP-2). About 70 different mutations of the V₂ receptor have been reported, most of them missense mutations. The functionally characterized mutants show a loss of function due to defects in their synthesis, processing, intracellular transport, AVP binding, or interaction with the G protein/adenylyl cyclase system. Thirteen different mutations of the AQP-2 gene have been reported. Functional studies of three AQP-2 mutations reveal impaired cellular routing as the main defect. The great number of different mutations with various functional defects hinders the development of a specific therapy. Gene therapy may, however, eventually become applicable to the congenital forms of NDI. At present all gene-therapeutic approaches lack safety and efficiency, which is of particular relevance in a disease that is treatable by an adequate water intake. The progress with regard to the molecular basis of antidiuresis contributes to the understanding of acquired forms of NDI on a molecular level. Recent data show that lithium dramatically reduces the expression of AQP-2. Likewise, hypokalemia reduces the expression of this water channel. The exact mechanisms leading to this reduced expression of AQP-2 remain to be determined.