Using data from 1998, we show that the gender log wage gap in Sweden increases throughout the wage distribution and accelerates in the upper tail of the distribution, which we interpret as a glass ceiling effect. Using earlier data, we show that the same pattern held at the beginning of the 1990?s but not in the prior two decades. Further, we do not find this pattern either for the log wage gap between immigrants and non-immigrants in the Swedish labor market or for the gender gap in the U.S. labor market. Our findings suggest that a gender-specific mechanism in the Swedish labor market hinders women from reaching the top of the wage distribution. Using quantile regressions, we examine whether this pattern can be ascribed primarily to gender differences in labor market characteristics or to gender differences in rewards to those characteristics. We estimate pooled quantile regressions with gender dummies, as well as separate quantile regressions by gender, and we carry out a decomposition analysis in the spirit of the Oaxaca-Blinder technique. Even after extensive controls for gender differences in age, education (both level and field), sector, industry, and occupation, we find that the glass ceiling effect we see in the raw data persists to a considerable extent.