Previous experience has shown that a noninvasive (indirect) technique using an oscillomet-ric monitor in conjunction with a tail cuff makes routine clinical blood pressure measurement practicable in dogs. The relationship between indirect and direct readings has been evaluated in both anaesthetised and conscious dogs (Bodey and others 1994, 1996). In this study, more than 2000 pressure measurements were taken from 1903 dogs. It was found that systolic is the most variable pressure parameter and that it depends on age, breed, sex, temperament, disease state, exercise regime and, to a minor extent, diet. Diet was not a significant determinant of diastolic and mean arterial pressure. Age and breed were the major predictors for all parameters. Heart rate was primarily affected by the temperament of the animal, though other factors also play a part in prediction. The distribution of systolic, diastolic, mean arterial pressure and heart rate across the dog population approximates to a log normal distribution. On the basis of these results it is possible to describe normal ranges for canine blood pressure; definition of hypertension, though, demands attention to age and breed normal values. The existence of statistically defined hypertension in an individual or breed does not imply adverse effects justifying therapy. Among the secondary causes of hypertension, such as diabetes, obesity and hyperadrenocorticism, hepatic disease was a new addition also undocumented in humans. The hypothesis that dogs, though classic model animals for hypertension, are resistant to its development found support from the modest increase in mean pressure values observed among dogs with renal disease, notably those with substantial reduction of glomerular filtration rate. The existence of breeds such as deerhounds with average pressures in the borderline range for hypertension in humans (and many individuals, therefore, well above) suggests that dogs may also be resistant to some of the adverse effects of high blood pressure.