The debate between Marshall and Jacobs theories on the economic performance of regions is complex, with empirical evidence showing mixed results. This paper reviews scholarly contributions and summarizes findings according to similarities and differences. The reviewed studies show a diverse picture of conditions under which each type of externality (specialisation or diversity) could be at work. The lack of resolution in the debate is generally not due to differences in the strength of agglomeration forces across industries, countries, or time periods, but rather due to measurement and methodological issues. The level of industrial and geographical aggregation, along with the choice of performance measures, specialisation and diversity indicators, are the main causes for the lack of resolution in the debate. The three-digit industrial classification seems to be the level at which MAR and Jacobs effects are undistinguishable from one another, and this is often exacerbated by a high level of geographical aggregation. The paper aims to provide a census of the papers that have dealt with the MAR-Jacobs dichotomy and to identify the threshold at which either theory becomes dominant from the point of view of the level of industrial aggregation, of spatial agglomeration, and so on. The paper also investigates why the literature remains relatively inconclusive. It finds that the results are often inconsistent, and that the exact spillover mechanism is not yet fully understood. The paper concludes that both specialized and diversified local industrial structures may promote economic performance of regions. The findings suggest that if regional specialisation may hinder economic growth, diversification is much less likely to induce this negative effect. The paper also discusses the role of externalities in economic growth and innovation, and the impact of different indicators and methodologies on the results. The paper concludes that the debate remains unresolved due to methodological issues, particularly the choice of independent variables.The debate between Marshall and Jacobs theories on the economic performance of regions is complex, with empirical evidence showing mixed results. This paper reviews scholarly contributions and summarizes findings according to similarities and differences. The reviewed studies show a diverse picture of conditions under which each type of externality (specialisation or diversity) could be at work. The lack of resolution in the debate is generally not due to differences in the strength of agglomeration forces across industries, countries, or time periods, but rather due to measurement and methodological issues. The level of industrial and geographical aggregation, along with the choice of performance measures, specialisation and diversity indicators, are the main causes for the lack of resolution in the debate. The three-digit industrial classification seems to be the level at which MAR and Jacobs effects are undistinguishable from one another, and this is often exacerbated by a high level of geographical aggregation. The paper aims to provide a census of the papers that have dealt with the MAR-Jacobs dichotomy and to identify the threshold at which either theory becomes dominant from the point of view of the level of industrial aggregation, of spatial agglomeration, and so on. The paper also investigates why the literature remains relatively inconclusive. It finds that the results are often inconsistent, and that the exact spillover mechanism is not yet fully understood. The paper concludes that both specialized and diversified local industrial structures may promote economic performance of regions. The findings suggest that if regional specialisation may hinder economic growth, diversification is much less likely to induce this negative effect. The paper also discusses the role of externalities in economic growth and innovation, and the impact of different indicators and methodologies on the results. The paper concludes that the debate remains unresolved due to methodological issues, particularly the choice of independent variables.