This paper studies the adjusted gender wage gap in Colombia at different points of the distribution of wages and the effect of sample selection. We compare 1994 with 2010. A glass ceiling pattern is observed after controlling for observed characteristics in both years. In addition, the sample selection is found to underestimate the adjusted gender wage gap. The effects of education, type of contract and type of employment on the gender wage gap across the wage distribution are also analyzed using a Dinardo, Fortin and Lemieux (1996) (DFL) type methodology. Educational change has helped to reduce the gender wage gap, mainly at the top of the wage distribution. By contrast, the reduction in the proportion of workers in the public sector and the reduction of workers with indefinite contracts have contributed to increasing the gender gap. The type of employment affected is mainly at the top of the distribution whereas the type of contract affected is at the bottom.