This article attempts to explore how key notions from Evolutionary Economics, such as selection, path-dependency, chance and increasing returns, may be applied to two key topics in Economic Geography. The first issue is the problem of how to specify the (potential) impact of the spatial environment on new variety in terms of technological change. Evolutionary thinking may be useful to describe and explain: (1) the process of localized `collective' learning in a regional context, (2) the adjustment problems that regions may be confronted with in a world of increasing variation, and (3) the spatial formation of newly emerging industries as an evolutionary process, in which the spatial connotation of increasing returns (that is, agglomeration economies) may result in a spatial lock-in. The second issue is the problem of how new variety may affect the long-term evolution of the spatial system. We distinguish three approaches that, each in a different way, apply evolutionary notions to the nature of spatial evolution. This is strongly related to the issue whether mechanisms of chance and increasing returns, rather than selection and path-dependency, lay at the root of the spatial evolution of new technology.