Conventional estimation methods on analyzing gender pay gap focus on comparing the earnings premium and gender inequality from the view of mean earnings distribution, highlighting human capital factors (e.g. education attainment, career training). However, mean distribution analysis do not reflect the whole perspective of gender earnings. Therefore, in our study, we adopt quantile regression estimation method to measure the impact of human capital (e.g. returns to education) and other social characteristics factors on wage. In addition, Melly2006 wage decomposition method is employed to reveal the pattern of gender earnings gap through overall distributions. We verified the evidence of ‘glass ceiling effect’ phenomenon in Korean labor market. The finds of our study also imply the female’s returns to education are higher than male, and the magnitude is even higher for upper earnings distribution. Furthermore, the estimation results of conditional and unconditional quantile regression present the differential of human capital variables occupy a big part of the explanatory on gender wage gap.