Nalmefene-loaded poly(lactic-co-glycolic acid) microspheres were prepared by O/O emulsification/solvent evaporation method. The central composite design-response surface methodology was used to optimize and predict the preparation microspheres. Effects of three independent variable variables i.e., Span80 concentration in outer phase, poly(lactic-co-glycolic acid) concentration in inner phase and theoretical drug content were evaluated on a number of response variables. Response variables selected in this study were drug content, encapsulation efficiency, mean diameter, diameter span and the cumulative percentage of the drug released in the first day after incubation (marked as F1d, and it was also calculated as the initial burst). Multiple linear regression and second-order polynomial model were fitted to the data, and the resulting equations were used to produce five dimensional response graphs, by which optimal experimental conditions were selected. The results showed that all response variables were greatly dependent on three independent variables, and the optimal conditions were Span80 concentration 1.5%, poly(lactic-co-glycolic acid) concentration 17.5%, and theoretical drug content 6%. According to the optimal conditions, the drug content, encapsulation efficiency, mean diameter, diameter span and F1d of prepared microspheres were 4.37%, 72.8%, 64.1 microm, 1.36 and 8.93%, respectively.