The relationship between the dose of intravenously administered streptozotocin (a N-nitroso derivative of glucosamine) and the diabetogenic response has been explored by use of the following indices of diabetogenic action: serum glucose, urine volume, and glycosuria, ketonuria, serum immunoreactive insulin (IRI), and pancreatic IRI content. Diabetogenic activity could be demonstrated between the doses of 25 and 100 mg/kg, all indices used showing some degree of correlation with the dose administered. Ketonuria was only seen with the largest dose, 100 mg/kg. The most striking and precise correlation was that between the dose and the pancreatic IRI content 24 hr after administration of the drug, and it is suggested that this represents a convenient test system either for both related and unrelated beta cytotoxic compounds or for screening for modifying agents or antidiabetic substances of a novel type. Ability to produce graded depletion of pancreatic IRI storage capacity led to an analysis of the relationship between pancreatic IRI content and deranged carbohydrate metabolism. Abnormal glucose tolerance and insulin response were seen when pancreatic IRI was depleted by about one-third, while fasting hyperglycemia and gross glycosuria occurred when the depletion had reached two-thirds and three-quarters, respectively. The mild yet persistent anomaly produced by the lowest effective streptozotocin dose, 25 mg/kg, exhibits characteristics resembling the state of chemical diabetes in humans and might thus warrant further study as a possible model. Finally, the loss of the diabetogenic action of streptozotocin by pretreatment with nicotinamide was confirmed and was shown to be a function of the relative doses of nicotinamide and streptozotocin and of the interval between injections.