Despite considerable progress in the development of immunoisolation devices, the optimal permeability of such devices is not known. This limitation stems partly from deficits in knowledge about which molecules should be allowed to traverse the semipermeable membrane and which molecules should be excluded, and also partly from experimental obstacles that have prevented a systematic study of permeability. To determine the optimal permeability of immunoisolation devices, we have created a series of microcapsules (800 microM diameter) that span a broad range of molecular exclusion limits yet are identical in wall thickness and chemical composition. Capsule permeability was precisely defined by two complementary methods--size exclusion chromatography (SEC) and a newly developed methodology to assess permeability of biologically relevant proteins. The entry of interleukin-1 beta-125I was significantly delayed, but not prevented, when the capsule exclusion limit was decreased from 230 kD to 3.2 kD, as determined by SEC with dextran standards. The influx of IgG was as predicted, based on the viscosity radius R eta of IgG and the capsule exclusion limit defined by SEC. Glucose-stimulated insulin secretion by encapsulated pancreatic islets did not differ as capsule permeability was decreased from a molecular exclusion limit of 230 kD to 120 kD. These studies should assist in the design of immunoisolation devices by defining the permeability optimal for cell function and also should be applicable to any cell type or immunoisolation device.