We revisit the nonconsensual econometric works – although the natural resource curse may have flourished – on the relationship between natural resources and economic performance. We first question the two terms of the relationship. We consider the role of institutions (separately and in interaction with the variable of interest) and of a number of usual or new control variables (income inequality and current account). The model, based on development accounting, is tested using four econometric techniques on the full sample (130 countries, 1990–2019) and by sub-samples according to per capita income, illustrating the non-linearity of the relationship. Three stylized facts emerge: first, the overall results converge towards a strong blessing of resource rents on GDP per capita. This can be explained mainly by the role of these rents in countries with very high GDP per capita. Second, institutional variables significantly mitigate the negative effect or reinforce the positive effect of these resources on development. Finally, among the categories of resources considered, it is the oil rent that favors this strong natural resource blessing. The effects of the observed categories may offset each other. Detailed analyses of estimation’s results in sub-samples and articulated with the results of the full sample are also proposed.