OBJECTIVES:  Our earlier study compared retrospectively health-care resources consumed 12 months before and 24 months after paliperidone long acting injection (PLAI) treatment in 60 patients with schizophrenic disorders from 2 public hospitals in Hong Kong. Results indicated that hospitalization was the major cost driver and substantial reduction was possible after PLAI therapy. The present study is to predict the potential annual savings after PLAI treatment by using data from 100 patients studied from Jan 2011 to Dec 2014. The study was performed from a government’s perspective. METHODS:  Predicted outcomes were generated using a multivariate generalized linear model (GLM) based on generalised estimating equations with identity and logit link functions with patients’ baseline characteristics, e.g. gender, income, marital status, occupation, conversion dosage and tapering effect adjusted as covariates. The main outcome measures were number of hospitalization, all-cause hospitalization, hospitalization due to schizophrenia. Other secondary outcomes such as emergency visit and compliance were also compared. RESULTS:  Compared to those receiving conventional antipsychotics, the GLM predicted (as per patient per year) a reduction in number of hospitalization by 0.96 episode, all-cause hospitalization by 7.78 days, and hospitalization due to schizophrenia by 7.47 days. In terms of safety profile, emergency visit was reduced by 1.489 visits after patients were switched to paliperidone. Based on the cost of hospitalization and emergency room visit by the government, this can potentially translate into a yearly savings of USD31,019 per patient. In addition, our study also demonstrates that the risk of non-compliance rate was substantially reduced by 75% which can lead to further health care savings. CONCLUSIONS:  Results of the present study suggests that PLAI therapy is likely to lead to long term savings by reducing hospitalization. The findings from our study may be useful in health-care decision making in treatment options for schizophrenia in resource-limited settings.