The aim of this study was to prepare two types of injectable cationized microspheres based on a native gelatin (NGMS) and aminated gelatin with ethylenediamine (CGMS) to prolong the action of insulin. Release of rhodamin B isothiocyanate insulin from CGMS was compared with that from NGMS under in-vitro and in-vivo conditions. Lower release of insulin from CGMS compared with that from NGMS was caused by the suppression of initial release. The disappearance of 125I-insulin from the injection site after intramuscular administration by NGMS and CGMS had a biphasic profile in mice. Almost all the 125I-insulin had disappeared from the injection site one day after administration by NGMS. The remaining insulin at the injection site after administration by CGMS was prolonged, with approximately 59% remaining after one day and 16% after 14 days. The disappearance of CGMS from the injection site was lower than that of NGMS. However, the difference in these disappearance rates was not great compared with those of 125I-insulin from the injection site by NGMS and CGMS. The time course of disappearance of 125I-CGMS from the injection site was similar to that of 125I-insulin by CGMS. The initial hypoglycaemic effect was observed 1 h after administration of insulin by NGMS, thereafter its effect rapidly disappeared. The hypoglycaemic effect was observed 2-4 h after administration by CGMS and continued to be exhibited for 7 days. The prolonged hypoglycaemic action by CGMS depended on the time profiles of the disappearance of insulin from muscular tissues, which occurs due to the enzymatic degradation of CGMS.