Schizophrenia, a psychiatric disorder, requires long-term treatment; however, large fluctuations in blood drug concentration increase the risk of adverse reactions. We prepared a long-term risperidone (RIS) implantation system that can stabilize RIS release and established in-vitro and in-vivo evaluation systems. Cumulative release, drug loading, and entrapment efficiency were used as evaluation indicators to evaluate the effects of different pore formers, polymer ratios, porogen concentrations, and oil–water ratios on a RIS implant (RIS-IM). We also built a mathematical model to identify the optimized formulation by stepwise regression. We also assessed the crystalline changes, residual solvents, solubility and stability after sterilization, in-vivo polymer degradation, pharmacokinetics, and tissue inflammation in the case of the optimized formulation. The surface of the optimized RIS microspheres was small and hollow with 134.4 ± 3.5 µm particle size, 1.60 SPAN, 46.7% ± 2.3% implant drug loading, and 93.4% entrapment efficiency. The in-vitro dissolution behavior of RIS-IM had zero-order kinetics and stable blood concentration; no lag time was released for over three months. Furthermore, the RIS-IM was not only non-irritating to tissues but also had good biocompatibility and product stability. Long-acting RIS-IMs with microspheres and film coatings can provide a new avenue for treating schizophrenia.