In an attempt to explain the male-female wage differential, we estimated the relative marginal productivity and relative wage of female workers compared to those of male workers, using panel data from Japanese firms. The estimation results indicate that firms hiring 10 percentage points more women produce 0.8 percent more given the total wage bill and other inputs. Crosssectional estimates that neglect firm fixed effects indicate that female workers' marginal productivity is 45 percent of male workers', while female wage is 30 percent of male wage. These estimates indicate that part of the wage differential cannot be explained by the productivity differential. The estimation that allows for the correlated productivity/demand shocks suggests the robustness of the results. The IV estimator that allows for firm-level fixed effects seems to suffer from the bias due to the positive correlation between productivity/demand shocks and female employee proportion. Evidence found in this study rejects the null hypothesis that the male-female wage differential reflects the male-female productivity differentials.