In moderately diabetic rats (plasma glucose 20-30 mmol/L), where there is some residual pancreatic islet function, normoglycemia can be restored by transplantation of pancreatic islets into the liver via the portal vein. To examine whether normoglycemia can also be achieved in more severely diabetic animals (which more closely resemble human type I diabetes), we have compared the effect of transplanting 1000 islets intraportally in Lewis rats made moderately diabetic (55 mg/kg streptozotocin injected IP while nonfasting) or severely diabetic (65 mg/kg streptozotocin injected IP while fasting). In the moderately diabetic rats in which residual pancreatic insulin was 128 +/- 40 mU insulin (2.0% of control), plasma glucose stabilized (32 +/- 2.8 mmol/L at 1 week, 34 +/- 2 mmol/L at 3 weeks) as did body weight (falling from 290 +/- 5 to 265 +/- 5 g at 1 week and 253 +/- 6 g at 3 weeks). In contrast, in severely diabetic rats in which residual pancreatic insulin was only 13.5 +/- 4.2 mU insulin (0.21% of control), there was a progressive rise in plasma glucose (30 +/- 1.3 mmol/L at 1 week, 49 +/- 4 mmol/L at 2 weeks, and 67 +/- 7 mmol/L at 3 weeks) and a progressive fall in body weight (from 304 +/- 10 to 260 +/- 5 g by week 1 and to 209 +/- 6 g by week 3). Following islet transplantation, nonfasting plasma glucose normalized in moderately diabetic rats (10.5 +/- 0.6 vs. 9.1 +/- 0.6 mmol/L in nondiabetic controls, NS) after 23 +/- 5 days. In contrast, in the severely diabetic rats plasma glucose stabilized at 32 +/- 5 mmol/L (p < 0.05 compared to moderately diabetic group) but did not normalize. This difference was not attributable to different plasma glucose levels at the time of transplantation (35.1 +/- 1.8 in moderately diabetic vs. 32.5 +/- 2.5 mmol/L in severely diabetic rats). These observations demonstrate that residual native beta-cells (equivalent to only 60-80 islets) contribute to the survival or function of intraportally transplanted islets.