We use quantile regression and counterfactual decomposition methods to explore gender gaps across the earning distribution for a sample of full-time employees in the Australian private sector. Significant evidence of a self-selection effect for the women into full-time employment (or of components of self-selection related to observable or unobservable characteristics) is, interestingly, not found to be relevant in the Australian context. Substantial gender earnings gaps (and glass ceilings) are established, however, with these earnings gaps found to be predominantly related to the women receiving lower returns to their observable characteristics than the men.