Introduction: Islet transplantation has become a promising treatment for type I diabetes mellitus due to recent success since the development of the Edmonton Protocol. Islet culture prior to transplantation is standard practice in several clinical islet programs. High-density islet culture is desirable because it reduces space and handling requirements during culture, but it exacerbates oxygen limitations and causing a reduction in islet viability. We investigated the effect of tissue density on total tissue recovery, viable tissue recovery, and tissue purity for conventional normoxic culture on a polystyrene dish. To improve islet quality in high density culture, we explored use of elevated pO2 or culture on an oxygen-permeable silicone rubber membrane. We applied a theoretical O2 transport model to investigate how O2 transport changes for each culture condition and then compared our predictions to the experimental data to determine whether O 2 is limiting during high density culture using these new techniques. Methods: Human islets were cultured for 34-60 hr at densities varying from 20-5300 IE/cm2 on either 500-um silicone membranes or solid bottom dishes in a humidified incubator with 5% CO2, 19% or 56% oxygen at 37°C. The quantity, viability, and purity of tissue were quantified using nuclei counts, oxygen consumption rate (OCR) measurements, and insulin immunostaining. Theoretically predicted profiles of oxygen partial pressure and islet viability were obtained by solution of the species conservation equation with the finite element method. Results: Low density islet culture on a polystyrene dish resulted in tissue adherence, which decreased as density increased. The adherent tissue had a lower fraction of original insulinpositive cells collected from culture as compared to the non-adherent tissue. The combined non-adherent and adherent tissue collected from low density culture accounted for nearly 100% of the original viable tissue placed into culture and had purity similar to that for tissue prior to culture. With conventional culture, recovery of total OCR decreased sharply as viable tissue density increased. At all densities, the fraction of original viable tissue collected from culture was higher with culture on polystyrene dishes in 56% oxygen and even higher with normoxic culture on silicone rubber. Theoretical predictions were qualitatively similar to experimental results but in general over predicted the amount of viable tissue recovered. Conclusions: In low density culture all original viable tissue can be recovered as both free and adherent tissue. In high density culture, recovery of viable tissue (1) decreases as culture density increases on a polystyrene surface, (2) increases with increasing external pO2, and (3) increases substantially with culture on silicone rubber by removing O2 limitations.