Reliable high-recovery human islet storage would facilitate tissue matching, organ sharing, and immune manipulation of donor islets and prospective diabetic recipients. Collagenase-isolated, Ficoll-purified pancreatic islets (median 21,000, 15% of total islet yield) from eight cadaver pancreases were cultured in vitro for 24 h, equilibrated in three steps with dimethyl sulfoxide (DMSO) to a 2-M concentration, supercooled, nucleated, and cooled at 0.25 degree C/min to -40 degrees C before storage at -196 degrees C for 44.25 +/- 8.75 days. Rewarming at 200 degrees C/min and removal of DMSO with 0.75 M sucrose preceded 48 h of culture and retesting. Recovery postthaw by microscope count on duplicate aliquots was 94.2 +/- 3.5% of prefreeze counts and by triplicate assay of extractable insulin was 90.0 +/- 22.3% on day 0 and 74.1 +/- 12.6% after a 48-h culture. Nonfrozen islets increased basal insulin secretion 7.7 +/- 2.8 times after stimulation with 300 mg/dl glucose in perifusion, whereas islets frozen-thawed and cultured 48 h increased 6.2 +/- 0.8 times (NS). Peak stimulated insulin release was 0.92 +/- 0.14 microU.islet-1.min-1 before storage and 0.73 +/- 0.14 microU.islet-1.min-1 (79% of control, NS) after freeze-thaw and a 48-h culture. Total insulin secretion (area under curve) was 66% of prefreeze values at 48 h. Immunocytochemical stains revealed preservation of islet morphology postthaw. Electron microscopy showed intact cellular and nuclear membranes and intracellular organelles. Frozen-thawed islets harvested 14 days after renal subcapsular xenografting in nude mice were revascularized and well granulated. Cryopreservation can achieve prolonged storage of large numbers of human islets with high recovery numerically and functionally, making this a feasible approach for future trials of human islet transplantation.