A critical element of a firm's technology strategy concerns the allocation of R&D resources across distinct technologies and product markets. Yet, beyond the idea of economies of scope, there is little predictive theory about how the extent and specific directions of R&D diversification may influence R&D performance. In this paper, we take an information-based approach to the problem. Following others, we view R&D as an information-processing activity geared toward technical problem-solving and R&D capabilities as residing in the firm's specific information-processing routines. We define relatedness of R&D fields in terms of similarities or differences in information processing problems faced by decision-makers in the firm. Projects with similar information processing problems are characterized as being within the same "information regime." Using project-level data from the pharmaceutical industry, we explore two questions: 1) Do firms tend to focus their R&D activities within similar information regimes? 2) Are there performance benefits of such focus? We find evidence that firms diversify their R&D projects portfolios across different information regimes. We also find that there are performance costs of such diversification.