To evaluate the physiological effects and toxicity of epidural clonidine.HCl, male Beagle dogs were prepared with chronic lumbar epidural catheters and administered constant infusions of either saline (N = 10), or 80 micrograms/hr (N = 6), 200 micrograms/hr (N = 6), or 320 micrograms/hr (N = 12) clonidine.HCl at a rate of 4 ml/24 hr for 28 days. Saline infusion had no effect upon any behavioral measure. Epidural clonidine produced a dose-dependent increase in thermal skin-twitch response latency (antinociception), lowering of respiration rate, heart rate, and blood pressure, and increased sedation. The effects were maximum from approximately Day 1 to Day 3 when, with the exception of respiration which remained depressed, a progressive adaptation was observed over the course of the study. There were no negative effects on body weight, body temperature, motor function, bowel or bladder function, or clinical pathology values. After 28 days of continuous infusion, the dogs were deeply anesthetized and terminated. Cisternal cerebrospinal fluid taken at termination displayed no clinically significant differences in protein or glucose concentration. All groups, including control, had dogs which had a chronic inflammatory response in the epidural space, as represented by fibrosis, foreign body giant cells, and lymphocytes, but no spinal cord pathology. Both the steady-state plasma and CSF concentrations of clonidine were proportional to the dose; the ratio of CSF to plasma concentration was approximately 0.5. The failure to see any change in CSF composition, significant spinal cord pathology, or signs of tissue or organ toxicity emphasizes the safety of epidurally administered clonidine at infusion rates up to 320 micrograms/hr and at infusate concentrations up to 2 mg/ml.