The knowledge base concept in the past has often been applied in its "pure" form, i.e. it was assumed that there are dominant knowledge bases in particular sectors and firms, that shape the knowledge- and innovation process and related networks. For example, it has been argued that in the case of "analytical sectors" such as biotech codified knowledge generated by universities and R&D organisations are key for radical innovation, and that such knowledge is often transmitted by formal R&D cooperations and university-firm links. "Synthetic sectors" such as machinery on the other hand were assumed to innovate more incrementally by recombining existing knowledge that was often drawn from suppliers or service firms. Empirical literature partly has confirmed these basic patters, but also has demonstrated that more complex knowledge processes are underlying these overly schematic expectations. In addition, there have been arguments by Asheim, Boschma and Strambach, among others, more recently that combinations of different but related knowledge bases and -assets might be of high relevance for understanding innovation processes of firms in particular sectors and regions. This implies that innovation of firms e.g. in "analytical sectors" might benefit not just from new and basic knowledge generated by research, but also from recombining existing and applied knowledge or by drawing on symbolic knowledge assets. The same argument for the relevance of combinatorial knowledge bases applies for "synthetic" and "symbolic sectors", but in different forms. The paper investigages if the reliance on combinatorial knowledge bases leads to a better innovation performance and more radical forms of innovation than the use of more narrow knowledge assets. The paper investigates the relevance of combinatorial knowledge bases for innovation at first conceptually and based on respective literature. In the second part we analyse this question empirically by drawing on findings for the ICT sector in three regions of Austria (Vienna, Upper Austria, and Salzburg). (authors' abstract) / Series: SRE - Discussion Papers