Bioassay of glucagon based on the hyper-glycaemic activity was investigated in normal rabbits. — It was shown that the index of precision (λ) derived from the dose response curve is about 1.2, which means that several hundred animals are required per assay in order to obtain a reasonably narrow confidence interval. Triple cross-over design improved the index of precision to 0.37, about 50 rabbits then being required to obtain a 95% confidence interval of ± 35%. The weight and sex of animals, as well as the fasting time and the route of administration of glucagon, were shown not to influence the quality of the assay. — Pretreatment of rabbits with cortisone improves the blood sugar response to glucagon. With a single dose of 25 mg of cortisone per rabbit, a considerable increase in the response to glucagon was observed with only a moderate elevation of the fasting blood sugar level. It was shown that the cortisone potentiation of the glucagon effect is at a maximum around 2–4 days after the cortisone adminstration, and that a cross-over assay can be performed within this period. — Further, it was shown that the cortisone treatment of the animals can be repeated several times without causing any changes in the glucagon response or in the animals as such. — Variables in an assay, such as sampling time, dose ratio, and response calculation, have been evaluated. — Based on these results, an assay procedure for the determination of the biological potency of glucagon employing cortisonepretreated rabbits has been designed. — With this procedure, a 95% confidence interval of ±20% is ordinarily obtained in assays employing 32 rabbits.