We investigated the effect of alginate-polylysine-alginate microencapsulation on glucose-induced insulin secretion by rat islets. Applying the encapsulation method originally described by Lim, we found severely reduced in vitro insulin release (expressed as picomoles of insulin.10 islets-1.45 min-1 when incubated in 16.5 mM glucose), because the insulin release with encapsulated islets was 1.42 +/- 0.49 compared to 13.58 +/- 0.80 with free control islets. This could not be explained by inadequate permeability of the capsule, because insulin release was also severely reduced (2.12 +/- 0.61) when islets were subjected to the procedure but without the membrane-forming polylysine step. Therefore, islets were tested after having been subjected separately to each of the steps of the procedure. Insulin release was not affected by either alginate or CaCl2 but was severely reduced after prolonged suspension in saline or treatment with citrate. When saline and citrate were replaced by Ca2(+)-free Krebs-Ringer bicarbonate buffer (KRBB) and 1 mM EGTA, respectively, insulin release improved significantly both with complete and with incomplete (no polylysine step) encapsulation. This outcome was verified in a set of experiments run in parallel with islets derived from the same isolation procedure. Insulin release was 1.20 +/- 0.23 from islets encapsulated with the method of Lim and 10.73 +/- 1.04 from free control islets. With the modified procedure, insulin release was 9.17 +/- 0.52 vs. 9.61 +/- 1.27 for complete versus incomplete encapsulation, respectively. We conclude that Ca2(+)-free KRBB instead of saline and EGTA instead of citrate should be used to obtain an adequate insulin response from encapsulated islets and that the capsule membrane as such has no influence on glucose and insulin diffusion.