This paper revises the thesis that exporting firms learn to be more innovative and efficient as they have contact with certain information flows from their foreign activity (e.g., from buyers, suppliers or competitors). The paper begins by exploring the connections between two distinct concepts: Self-Selection (of more efficient firms into exports) and Learning-by-Exporting. The study then proceeds with a comparative analysis of the most recent literature and presents common facts and evidence, as well as key issues still open to debate. Learning-by-Exporting should be measured directly using firms´ innovative performance. However, given the lack of suitable data on firms’ innovative activities most studies have followed an indirect approach, using productivity measures. Several methodologies have been employed to estimate Total Factor Productivity and to test the Learning-by-Exporting hypothesis, but so far no final consensus has been reached on the best way to do it.