This study examines whether the loss of metabolic control in initially normalized islet transplants can result from the inadequate composition of the donor tissue. Streptozotocin-induced diabetic rats were followed for 64 weeks after the intraportal injection of islet isografts with different composition. The implantation of 2.3 million beta-cells (10(7)/kg body wt) as particles (>100 microm diameter) of primarily insulin-positive (70%) and glucagon-positive (20%) cells succeeded in a long-term normalization of 2-h fasting glycemia, glucose tolerance, and serum fructosamine. The same metabolic control was achieved in animals with short and long durations of diabetes or when grafts were implanted under the kidney capsule. At posttransplantation (PT) week 64, insulin reserves were 60% lower than those in age-matched controls, which may account for the glucose intolerance in a few old recipients. The same type of graft containing 0.7 million beta-cells (4 x 10(6)/kg body wt) corrected these metabolic parameters for more than 12 weeks; the proportionally lower insulin reserves were sufficient for the long-term correction of 2-h fasting glycemia, but did not avoid glucose intolerance in older recipients. When the higher beta-cells number (10(7)/kg body wt) was injected as smaller particles (<100 mpm diameter) of lower purity (55% insulin-positive) and negligible glucagon content (<5% glucagon-positive), the metabolic parameters were also corrected for 12 weeks PT but then progressively returned to overt diabetes (6 of 10) or glucose intolerance (4 of 10). We concluded that long-term metabolic normalization can be achieved by islet implants in the liver or under the kidney capsule. The loss of metabolic control in older animals can be caused by the inadequate composition of the graft, with the number of beta-cells, the proportion of other endocrine and nonendocrine cells, and the particle size as influential variables.