This paper contributes the economics of knowledge and innovation with the analysis of the knowledge cost function and sheds light on the determinants of the large variance in the cost of innovation across firms. The amount and the structure of external knowledge and the internal stocks of knowledge that firms can access and use in the generation of new technological knowledge help firms to reduce the costs of innovations. The empirical section is based upon a panel of companies listed on UK and the main continental Europe financial markets (Germany, France and Italy) for the period 1995 – 2006, for which information about patents have been gathered. The econometric analysis of the costs of knowledge considers the unit costs of patents on the right hand side, and on the left hand side next to R&D expenditures, the stock of knowledge internal and external to each firm. In order to articulate the different facets of the external knowledge that is made accessible by proximity with firms co-localized in the same region (NUTS2), we further include other variables proxying for regional variety, complementarity and similarity. The results confirm that the stock of internal knowledge and the access to external knowledge play a key role in reducing the actual cost of the generation of new technological knowledge at the firm level.