At the same time as we can observe strong tendencies of a globalisation of R&D, we also can observe a strong spatial clustering of R&D and related innovative activities. The standard explanation in the literature of the clustering of innovative activities is that such clusters offer external knowledge economies to innovative companies, since they are dependent upon knowledge flows and that knowledge flows are spatially bounded. Obviously, location is crucial in understanding knowledge flows and knowledge production, since knowledge sources have been found to be geographically concentrated. There are two major performers of R&D: industry and universities. It seems rather straight-forward to assume that industrial R&D might be attracted to locate near research universities doing R&D in fields relevant to industry. Already as far back as in the 1960s a number of case studies confirmed the important roles played by Stanford University and MIT for commercial innovation and entrepreneurship. During the years a large number of formal studies have presented evidences of a positive impact of university R&D on firm performance. The question is, does it also work the other way around? Does industrial R&D function as an attractor for university R&D? We may actually think of several reasons why university R&D may grow close to industry R&D. First of all political decision-makers may decide to start or expand university R&D at locations where industry already is doing R&D. Secondly, we can imagine that industry doing R&D in a region might use part of their R&D funds to finance university R&D. Thirdly, universities in regions with industrial R&D might find it easier to attract R&D funds from national and international sources due to co-operation with industry. Obviously, not all types of university R&D attract industrial R&D. There are reasons to believe that, in particular, university R&D in natural, technical and medical sciences attracts industrial R&D but that there are also strong reasons to believe that there are variations between different sectors of industry regarding how dependent their R&D is to be located close to university R&D. The above implies that there are behavioural relationships between industrial R&D and university R&D and vice versa. However, the litrature contains few studies dealing with this problem. Most studies have concentrated on the one-directional effect from university R&D to industrial R&D and the outputs of industrial R&D in most cases measured in terms of the number of patents and neglected the possible mutual interaction. However, if there is a mutual interaction between university and industry R&D, and if there are knowledge externalities involved, then we can develop a dynamic explanation to the clustering of innovative activities based on positive feedback loops. This would imply strong tendencies to path dependency and that policy initiatives to transfer non-innovative regions to innovative regions would have small chances to succeed. The fact that knowledge flows seem to be spatially bounded implies that proximity matters. Most contributions analysing spatial knowledge flows have used very crude measures of proximity. However, there are some authors that have argued that proximity could be measured using accessibility measures. Accessibility measures can be used to model interaction opportunities at different spatial scales: local, intra-regional and inter-regional. The purpose of this paper is to analyse the locational relationship between industry R&D and university R&D in Sweden using a simultaneous equation approach and to analyse existing differences between different science areas and different industry sectors.