This study uses quantile regression models to examine whether the union wage effect varies across the conditional wage distribution. Although for men it is evident that the union wage effect decreases when moving up the conditional wage distribution, the effect for women is relatively stable except at the extremities of the distribution. Overall, unions are found to have a larger effect on men than on women wages. The decomposition results show that for men, the union wage effect explains a substantial proportion of the observed wage gap between union and non-union workers; this is not the case for women.