This paper argues that technological systems (defined as networks of agents interacting in each specific area of technology under a particular institutional infrastructure) constitute a useful unit of analysis for not only innovation and diffusion studies but also work related to technology policy. In the case of Sweden's technological system for factory automation (e.g. robots), it is suggested that three features are central to explaining the extensive diffusion of new technology in the Swedish engineering industry, namely 'bridging institutions', user-supplier linkages, and critical mass coupled to advanced economic competence among a key set of users. As regards public policy, it is argued that its role is to facilitate the formation of new technological systems and to enhance the functioning of existing systems rather than to rectify individual market failures. This can be done by creating or strengthening 'bridging institutions' and centers of excellence, promoting speedy and wide diffusion of engineering competence, and strengthening compensating mechanisms when important nodes in the networks fail. Given the difficulty of identifying future technological opportunities, the simultaneous pursuit of alternative strategies, early identification of important developments, and increasing the absorptive capacity of the economy become important aspects of public policy.