Productivity levels and productivity growth rates vary significantly over space. These differences are perhaps most pronounced between countries, but they remain acutely evident within national spaces as economic growth favors some cities and regions and not others. In this paper, we map the spatial variation in productivity levels across Canadian cities and we model the underlying determinants of that variation. We have two main goals. First, to confirm the existence, the nature and the size of agglomeration economies, that is, the gains in efficiency related to the spatial clustering of economic activity. We focus attention on the impacts of buyer-supplier networks, labour market pooling and knowledge spillovers. Second, we identify the geographical extent of knowledge spillovers using information on the location of individual manufacturing plants. Plant-level data developed by the Micro-economic Analysis Division of Statistics Canada underpin the analysis. After controlling for a series of plant and firm characteristics, analysis reveals that the productivity performance of plants is positively influenced by all three of Marshall's mechanisms of agglomeration (Marshall, 1920). The analysis also shows that the effect of knowledge spillovers on productivity is spatially circumscribed, extending, at most, only 10 km beyond individual plants. The reliance of individual businesses on place-based economies varies across the sectors to which the businesses are aggregated. These sectors are defined by the factors that influence the process of competition'access to natural resources, labour costs, scale economies, product differentiation, and the application of scientific knowledge. Neither labour market pooling, buyer-supplier networks nor knowledge spillovers are universally important across all sectors. This paper provides confirmation of the importance of agglomeration, while also providing evidence that external economies are spatially bounded and not universally important across all industries.