The paper revisits the relationship between GDP per capita and diversification, using classical and more recent trade theory. Three theoretical findings are presented: (i) competitive models yield predictions only for the extensive product margin; (ii) countries continuously diversify their production and exports—a major controversy in the empirical literature; and (iii) causality runs from diversification to GDP per capita, and not the other way around. The theoretical analysis also provides indication for the appropriateness of alternative measures of diversification, and enables estimating the relationship to economic development in a gravity-type parametric specification. Using detailed data on countries’ exports, the case of re-specialization is rejected. Inference of causality reveals some evidence for GDP per capita affecting the level of diversification, but stronger support for diversification affecting GDP per capita. Generally, both variables are highly endogenous as they are both driven by the technology parameters in standard models of economic growth and international trade.