A perifusion system, intermediate in character between perfusion and static incubation, is described for the investigation of the multiphasic aspects of insuiin secretion from pancreas not subjected to separation of endocrine and exocrine tissue. Pancreatic tissue fragments obtained by mechanical chopping with razor blades are placed in small (0.7 ml.) chambers through which medium is pumped and from which effluent is collected at various time intervals for IRI determination. Under the conditions employed, insulin degradation by released proteolytic enzymes is minimal throughout the test period. Dose-dependent glucose-induced insulin release is observed and decreases immediately after cessation of the stimulus. Pancreas from fed animals secretes significantly greater quantities of insulin in response to glucose thata pancreas from fasted animals. Optimal results are obtained when the chopped tissue fragments average (0.5 mm.)3. The secretion profiles obtained with this system are similar to those reported with pancreatic perfusion. Tolbutamide produced an early insulin response only, whereas glucose and amino acids induced both early and late insulin secretory responses. Among the hexoses tested, only glucose and mannose stimulated insulin secretion, whereas two inhibitors of the early steps of glucose metabolism, mannoheptulose and 2-deoxygIucose, inhibited both the early and late phases of glucose-induced insulin secretion. The separation between the two phases of insulin secretion is not as sharp and complete with perifusion as with perfusion. This difference is presently attributed to the greater role of tissue diffusion in the minced perifused tissue. Balancing this disadvantage, however, perifusion offers several advantages over perfusion, among which are ease of tissue preparaton and handling, ability to study a number of variables relatively rapidly, comparison of identically prepared and randomly sampled minces of the same pancreas in parallel. Finally, the preparation is applicable whenever pancreatic perfusion is not possible (for example, in the fetus or for very small animals), or when the isolation of islets of Langerhans presents practical or theoretical disadvantages, as in the study of the influences of exocrine on endocrine tissue.