The purpose of the present study was to determine the contributions of discrete insulin secretory bursts vs. basal insulin release to total insulin secretion in vivo. Quantification of the partitioning of pulsatile and basal insulin secretion is complicated by physiological delivery of these pulses into the portal vein and the absence of validated methods of measuring the rates of pulsatile and basal insulin secretion in vivo. We therefore 1) developed a canine model with chronically implanted portal vein catheters, 2) validated an established deconvolution technique as well as a novel direct catheterization technique (Clustcath) for measurement of pulsatile and nonpulsatile insulin secretion rates in this model, and 3) applied these methods to study insulin secretion in the overnight-fasted dog in vivo to determine the contribution of pulsatile vs. basal insulin secretion to total rates of endogenous insulin secretion. Rates of total, pulsatile, and nonpulsatile endogenous insulin secretion measured by Cluscath closely parallel those measured by deconvolution analysis (54 +/- 15 vs. 51 +/- 11, 38 +/- 12 vs. 36 +/- 11, and 16 +/- 4 vs. 14 +/- 4 pmol/min, respectively). Clustcath and deconvolution indicated that the majority of insulin was secreted as pulses (70 +/- 6 and 66 +/- 7%, respectively). These data infer that any process that selectively decreases the pulsatile component of insulin secretion (e.g., diabetes mellitus) will likely have a major impact on total insulin secretion.