This paper examines total factor productivity differences between exporting and non-exporting firms. These differences are documented on the basis of a sample of Spanish manufacturing firms over the period 1991–1996. The paper also examines two complementary explanations for the greater productivity of exporting firms: (1) the market selection hypothesis, and (2) the learning hypothesis. Non-parametric tests are proposed and implemented for testing these hypotheses. Results indicate clearly higher levels of productivity for exporting firms than for non-exporting firms. With respect to the relative merits of the selection and the learning hypotheses, we find evidence supporting the self-selection of more productive firms in the export market. The evidence in favor of learning-by-exporting is rather weak, and limited to younger exporters.