Dogs are well established as experimental animals for the study of both renal disease and hypertension, but most work is based on surgical or pharmacological models and relatively little on spontaneous diseases. This review argues for the latter as an underexploited aspect of comparative medicine. The most important feature of canine hypertension may not be the ease with which models can be produced but the fact that dogs are actually rather resistant to hypertension, and perhaps to its effects, even when they have chronic renal failure. The importance of natural models of chronic renal failure is strengthened by the evidence that self-sustaining progression is a consequence of extreme nephron loss, that is, a late event, rather than the dominant feature of the course of the disease. The role of salt in hypertension is discussed and emphasis given to the importance of understanding the physiological basis of nutritional requirement and recognizing that it is unlikely to exceed 0.6 mmol/kg/day for most healthy adult mammals except during pregnancy or lactation. Such a perspective is essential to the evaluation of experiments, whether in animals or humans, in order to avoid arbitrary definitions of 'high' or 'low' sodium intake, and the serious misinterpretations of data which result. An age-related rise in arterial pressure may well be a warning of excess salt intake, rather than a normal occurrence. Problems of defining hypertension in the face of variability of arterial pressure are also discussed.