One of the major steps toward successful islet transplantation for the treatment of type diabetes is to obtain islets of sufficient number and viability. Using a standardized method of isolating islets, the goal of this study was to analyze the factors influencing the outcome of islet isolation. A total of 104 cadaveric human pancreata were processed for islets by the same team. Data from the islet-processing charts were reviewed retrospectively. The two endpoints were the recovery of islets, viable after 2 days of culture (group V = viable, group NV = nonviable) and the islet yield. Viable islets were recovered in 61% of cases (n = 63). Minimal blood glucose recorded during hospitalization was very significantly lower in group V (124 +/- 5 vs. 148 +/- 9, P = 0.01). Lack of significant medical history in the donor was associated with better viability as compared with various donor predispositions (chi-2 4.21, P = 0.04). Cold ischemia time (8.1 +/- 0.5 hr in group V vs. 9.8 +/- 0.9 hr in group NV, P = 0.07) and collagenase lot (5 lots tested, chi-2 13.1, P = 0.01) also affected the recovery of viable islets. Hospital time was shorter in group V (65.3 +/- 6.8 vs. 80.9 +/- 17.9 hr, P = 0.35). Multivariate logistic regression analyses of viable islet recovery identified minimal blood glucose (P = 0.03) and collagenase lot (P = 0.06) as the most significant risk factors. However, the best multivariate predictive model--which includes blood glucose, collagenase lot, donor age and surgical procurement team--correctly predicted 66.2% of cases only. Multivariate analysis of final islet yield designed hospitalization length, cardiorespiratory arrest, surgical procurement team, and collagenase lot as the best predictors. These data obtained in a large series of pancreata emphasized several donor and technical factors that should target the attention of islet transplant researchers in order to improve islet yield and viability.