Oscillometric measurements of arterial blood pressure were compared with direct measurements made on seven dogs fitted with catheters. Tail and limb cuff sites were used while the dogs were gently restrained either standing or in lateral recumbency. The accuracy of the readings for the various cuff sites was compared with the direct (gold standard) readings. The accuracy of the indirect readings was improved by using mean values from a series of readings rather than individual values and when the dogs were in lateral recumbency rather than standing. The differences between the direct and indirect values were greatest with high pressures, and with systolic rather than diastolic values. In standing dogs, the proximal forelimb readings (when obtainable) correlated most closely with the direct readings. The tail cuff readings correlated significantly with the direct readings, though less closely. The tail cuff readings were the most easily recorded in the standing dogs. In the laterally recumbent dogs, the readings from all the cuff sites correlated closely with the direct values except for diastolic readings from the distal hindlimb.