In this study, a multiplicative semiparametric fourth order bias reduction density is proposed. The proposed density accommodates all parametric distributions and produces optimum results even in situations where the underlying parametric distributions are not the best approximations to the correct density of the data. This offers an impressive option for estimating statistics with large sample propertiessuch as complex indixes. It consists of parametric estimate multiplied by a nonparametric correction function. The simulation results show its practical potential