Poly(D,L-lactide-co-glycolide)(PLGA) microspheres were prepared by emulsion solvent evaporation method. The influences of inner aqueous phase, organic solvent, PLGA concentration on the morphology of microspheres were studied. The results showed that addition of porogen or surfactants to the inner aqueous phase, types of organic solvents and polymer concentration affected greatly the microsphere morphology. When dichloromethane was adopted as organic solvent, microspheres with porous structure were produced. When ethyl acetate served as organic solvent, two different morphologies were obtained. One was hollow microspheres with thin porous shell under a lower PLGA concentration, another was erythrocyte-like microspheres under a higher PLGA concentration. Three types of microspheres including porous, hollow core with thin porous shell(denoted by hollow in brief) and solid structures were finally selected for in vitro drug release tests. Bovine serum albumin(BSA) was chosen as model drug and encapsulated within the microspheres. The BSA encapsulation efficiency of porous, hollow and solid microspheres was respectively 90.4%, 79.8% and 0. And the ultimate accumulative release was respectively 74.5%, 58.9% and 0. The release rate of porous microspheres was much slower than that of hollow microspheres. The experiment results indicated that microspheres with different porous structures showed great potentials in controlling drug release behavior. The influences of inner aqueous phase, organic solvent, PLGA concentration on the morphology of microspheres were studied. The results showed that addition of porogen or surfactants to the inner aqueous phase, types of organic solvents and polymer concentration greatly greatly microsphere morphology. when ethyl was served as organic solvent, One was hollow microspheres with thin porous shell under a lower PLGA concentration, another was erythrocyte-like microspheres under a higher PLGA concentration. Three types of microspheres including porous hollow cores with thin porous shells (denoted by hollow in brief) and solid structures were finally selected for in vitro drug release tests. Bovine serum albumin (BSA) was chosen as model drug and encapsulated within the microspheres. The BSA encapsulation efficiency of porous, hollow and solid microspheres was respectively 90.4%, 79.8% and 0. And the ultimate accumulative release was respectively 74.5%, 58.9% and 0. The release rate of porous microspheres was much slower than that of hollow microspheres. The experiment results indicated that microspheres with different porous structures showed great potentials in controlling drug release behavior.