Funding agencies and large public scientific institutions are increasingly using the term “research portfolio” as a means of characterising their research. While portfolios have long been used as a heuristic for managing corporate R&D (i.e., R&D aimed at gaining tangible economic benefits), they remain ill-defined in a science policy context where research is aimed at achieving societal outcomes. In this article we analyze the discursive uses of the term “research portfolio” and propose some general considerations for their application in science policy. We explore the use of the term in private R&D and related scholarly literature in existing science policy practices, and seek insight in relevant literature in science policy scholarship. While the financial analogy can in some instances be instructive, a simple transposition from the world of finance or of corporate R&D to public research is problematic. However, we do identify potentially fruitful uses of portfolio analysis in science policy. In particular, our review suggests that the concept of research portfolio can indeed be a useful analytical instrument for tackling complex societal challenges. Specifically, the strands of scholarship identified suggest that the use of research portfolio should: i) recognize the diversity of research lines relevant for a given societal challenge, given the uncertainty and ambiguity of research outcomes; ii) examine the relationships between research options of a portfolio and the expected societal outcomes; and iii) adopt a systemic perspective to research portfolios – i.e., examine a portfolio as a functional whole, rather than as the sum of the its parts. We argue that with these considerations, portfolio-driven approaches may foster social inclusion in science policy decisions, help deliberation between “alternative” portfolios to tackle complex societal challenges, as well as promote cost-effectiveness and transparency.