A vast literature on the international activities of heterogeneous firms finds the existence of a positive exporter productivity premium. On average, exporting firms are more productive than firms that sell on the national market only. The Melitz (Econometrica 71:1695-1725, 2003) model, however, has implications for not only mean differences but also differences in the distribution of productivity. Furthermore, exporting firms may be different from non-exporting firms for reasons that are not included in the Melitz model. We believe that conditioning on firm fixed effects and studying the distribution of productivity are both necessary for empirical tests of the Melitz model. This paper is the first to employ a new quantile estimation technique for panel data introduced in Powell (Did the economic stimulus payments of 2008 reduce labor supply? Evidence from quantile panel data estimation. RAND Corporation Publications Department, Santa Monica, 2014). We find that the premium is positive at all productivity levels, but highest at the lowest quantiles. These results support theoretical models which suggest that there is a division in productivity between exporters and non-exporters.