In this paper we study two closely related issues. First, the role of technology heterogeneity and diffusion in the convergence of GDP per worker observed across the European regions, in the absence of data on regional TFP. Second, the spatial pattern of the observed regional heterogeneity in technology and the relevance of this pattern for the econometric analysis of regional convergence in Europe. As for the first issue, our aim is to assess whether the convergence observed across European regions is due to convergence in technology as well as to convergence in capital-labour ratios. We first develop a growth model where technology accumulation in lagging regions depends on their own propensity to innovate and on technology diffusion from the leading region, and convergence in GDP per worker is due to both capital deepening and catch-up. We use data (1978-97) on 131 European regions. Propensities to innovate are computed by assigning each patent collected by the European Patent Office to its region of origin (see www.crenos.it) . Our findings are consistent with the hypothesis that technology differs across regions and that convergence is partly due to technological catch-up. As for the second empirical issue, we study to what extent each region's propensity to innovate is correlated with that of the surrounding regions. Our results show, first, that the performance of each region does depend on that of the surrounding areas. Second, that the intensity of such spillovers fades with distance. Taken together, these findings suggest the existence of significant localised spillovers of technological knowledge. Finally, we show that these spillovers are strong enough to play a role that cannot be ignored in the econometric analysis of the convergence process in Europe.