Venous injection of alloxan monohydrate is a standard method to produce a canine model of diabetes. Others have reported mortalities greater than 45 per cent and yields of diabetic dogs of less than 36 per cent with this technique. In this study, a new method for alloxan diabetogenesis is reported upon: alloxan monohydrate is injected intravenously with protection of the renal arteries at the time of injection by a 7F, triple lumen double balloon catheter placed in the abdominal aorta. The balloons are inflated under fluoroscopic control to occlude the renal arteries at the time of injection. Forty-three age-matched beagle dogs were initially injected with 60 milligrams per kilogram of alloxan monohydrate: 26 or 61 per cent became diabetic-defined as persistently doubled fasting serum glucose and glucosuria; ten failed to become diabetic, 23 per cent, and seven died, 16 per cent. The ten initial failures were reinjected with 65 milligrams per kilogram of alloxan monohydrate: six or 60 per cent then became diabetic, three were persistent failures, 30 per cent, and one dog died, 10 per cent. Thus, the over-all yield of diabetic dogs was 74 per cent, with an 18 per cent mortality. Minimal renal damage occurred, as evidenced by creatinine clearance, blood urea nitrogen and renal biopsy studies. These results suggest a significantly improved method--a twofold improvement over standard success rates with a twofold less mortality--of producing diabetic dogs by alloxan injection.