We investigate different techniques to assess the gender pay gap in five EU countries (France, Germany, Italy, Spain and United Kingdom), focusing on self-selection into market work. Results show that selectivity correction has an impact on both wage estimates and wage gap decomposition. If there is a positive correlation between the wage and the propensity to participate, the estimated pay gap understates the true difference in earnings when self-selection is ignored. The estimated pay gap differs considerably at different quantiles of the wage distribution, and is sensitive to the choice of estimator.