The present study was aimed at developing a delivery system for controlled release of insulin at basal level from chitosan-zinc-insulin complex incorporated into thermosensitive polymer, poly(lactic acid)-poly(ethylene glycol)-poly(lactic acid) (PLA-PEG-PLA). Chitosan-zinc-insulin complex was optimized to restrict the insulin diffusion from the delivery system by complex formation and thereby reducing initial burst release. Polymer concentration, insulin loading, chitosan and Zinc<super>+2</super> addition were shown to affect the insulin release in vitro. Formulations containing insulin, zinc-insulin, and chitosan-insulin exhibited high initial burst (~7-14%), accompanied with a large secondary burst and incomplete release. Chitosan-zinc-insulin containing formulations showed extended release profiles of insulin for 84-90 days with a significant (P<0.05) reduction in initial burst release and minimal secondary burst. Increasing chitosan amount had no effect (P>0.05) on the initial burst, and release rate. Insulin alone and zinc-insulin containing formulations showed significant (p<0.05) attenuation in secondary and tertiary structure of insulin, as compared to chitosan-zinc-insulin. The complex formation conserved the physical and chemical stability of insulin and protected it from aggregation during release and storage. It also protected insulin from the acidic degradation product of copolymer. The delivery systems were investigated for continuous in vivo insulin delivery at basal level for prolonged period after a single subcutaneous injection. In vivo absorption and bioactivity of insulin released were studied in streptozotocin-induced diabetic rats. Chitosan-zinc-insulin complex significantly (P<0.05) reduced the initial burst release of insulin in comparison to zinc-insulin or insulin alone. The delivery system released insulin for ~70 days in biologically active form with corresponding reduction in blood glucose. Blood glucose levels were comparable to that of control for longer duration, and were significantly (P<0.05) lower than untreated diabetic animals. No significant difference (P>0.05) in blood glucose levels in two consecutive time points until 56-63 days indicated a pharmacodynamic manifestation of continuous release of insulin at steady rate. The delivery systems showed increase in bioavailability of insulin (1.2-2 fold increase in AUC) as compared to zinc-insulin and insulin alone. Insulin released from the delivery systems did not provoke any immune response. The delivery systems were biocompatible in vitro and in vivo and were non-toxic.