Abstract: In the present study, a risperidone loaded microsphere/sucrose acetate isobutyrate (SAIB) in situ forming complex depot was designed to reduce the burst release of SAIB in situ forming depot and to continuously release risperidone for a long-term period without lag-time. The model drug risperidone (Ris) was first encapsulated into microspheres and then the Ris-microspheres were embedded into SAIB depot to reduce the amount of dissolved drug in the depot. The effects of different types of microsphere matrix, including chitosan and poly(lactide-coglycolide) (PLGA), matrix/Ris ratios in microspheres and morphology of microspheres on the drug release behavior of complex depot were investigated. In comparison with the Ris- loaded SAIB depot (Ris-SAIB), the complex depot containing chitosan microspheres (in which chitosan/Ris = 1:1, w/w) (Ris-Cm-SAIB) decreased the burst release from 12.16% to 5.80%. However, increased drug release rate after 4 days was observed in Ris-Cm-SAIB, which was caused by the high penetration of the medium to Ris-Cm-SAIB due to the hydrophilie of chitosan. By encapsulation of risperidone in PLGA microspheres, most drugs can be prevented from dissolving in the depot and meanwhile the hydrophobic PLGA can reduce the media penetration effect on the depot. The complex depot containing PLGA microspheres (in which PLGA/ drug = 4:2, w/w) (Ris-Pm-SAIB) showed a significant effectiveness on reducing the burst release both in vitro and in vivo whereby only 0.64% drug was released on the first day in vitro and a low AUC_0-4d value (105.2 ± 24.4) ng·mL^-1·d was detected over the first 4 days in vivo. In addition, drug release from Ris-Pm-SAIB can be modified by varying the morphology of microspheres. The porous PLGA microspheres could be prepared by adding medium chain triglyceride (MCT) in the organic phase which served as pore agents during the preparation of PLGA microspheres. The complex depot containing porous PLGA microspheres (which were prepared by co-encapsulation of 20% MCT) (Ris-PPm-SAIB) exhibited a slightly increased AUC_0-4d of (194.6 ± 15.8) ng·mL^-1·d and high plasma concentration levels from 4 to 78 days C_s(4-78d) = (7.8 ± 1.2) ng·mL^-1. The plasma concentration on 78 day C_78d was (9.0 ± 2.5) ng·mL^-1 which was higher than that of Ris-Pm-SAIB C_78d = (1.6 ± 0.6) ng·mL^-1. In comparison with Ris-Pm-SAIB, the AUC_4-78d of Ris-PPm-SAIB increased from (379.0 ± 114.3) ng·mL^-1·d to (465.0 ± 149.2) ng·mL^-1·d, indicating sufficient drug release from the Ris-PPm- SAIB. These results demonstrate that the risperidone loaded porous PLGA microsphere/SAIB in situ forming complex depot could not only efficiently reduce the burst release of SAIB depot both in vitro and in vivo, but also release the drug sufficiently in vivo, and be capable to continuously release the drug for 78 days.