In this paper, we use quantile regression decomposition methods to analyze the gender gap between men and women who work full time in the Netherlands. Because the fraction of women working full time in the Netherlands is quite low, sample selection is a serious issue. In addition to shedding light on the sources of the gender gap in the Netherlands, we make two methodological contributions. First, we prove that the Machado-Mata quantile regression decomposition procedure yields consistent and asymptotically normal estimates of the quantiles of the counterfactual distribution that it is designed to simulate. Second, we show how the technique can be extended to account for selection. We find that there is a positive selection of women into full-time work in the Netherlands; i.e., women who get the greatest return to working full time do work full time. We find that about two thirds of this selection is due to observables such as education and experience with the remainder due to unobservables. Our decompositions show that the majority of the gender log wage gap is due to differences between men and women in returns to labor market characteristics rather than to differences in the characteristics. This is true across the wage distribution, particularly in the top half of the distribution.