Although it has been recognized for almost 70 years that there is a substantial genetic component to the pathogenesis of hypertension, only recently have systematic efforts been made to identify the responsible genetically determined mechanisms. In the case of several rare syndromes, spectacular progress has been made in identifying the underlying molecular mechanisms responsible for the clinical expression. Glucocorticoidsuppressible aldosteronism and Liddle's syndrome, each inherited as an autosomal-dominant condition, complete the list. In the case of randomly selected patients and families with essential hypertension, inheritance involves many genes and progress has been far more modest. Probably the most promising lead has involved the genes governing the structure of angiotensinogen, the substrate in the renin reaction. Linkage has been established and confirmed. At the moment, however, neither the relation of the genetic abnormality to the underlying mechanisms, nor the contribution of this abnormality to hypertension in the individual patient, has been defined. We know less about other candidate genes, with the exception of studies that rigorously ruled out a contribution. The development of the concept of the “intermediate phenotype,” a physiological feature that makes it possible to identify a homogeneous subpopulation, should help to sort out many of these issues. Unfortunately, the identification and characterization of intermediate phenotypes is substantially more difficult at the moment than are the genetic studies, and so progress is likely to be slow. The field is complicated by the reporting of claims made on the basis of small patient samples. In the case of polymorphisms in the angiotensin-converting enzyme gene as a risk factor for tissue injury, for example, substantial follow-up studies have systematically failed to confirm the original report, which was based on a small patient sample. The fact that the same DNA collection is likely to be examined many times for multiple gene candidates creates a setting in which type I errors are likely, and so we are likely to see many more examples. Caveat lector. Again, the development of relevant intermediate phenotypes will make the spurious association less likely.