Research on the financial performance outcomes of open innovation has been equivocal and often relies on cross-sectional data and problematic assumptions about the role of the external context. A longitudinal perspective is crucial for gaining a better understanding of the potential of decreasing innovation utility as well as the conditions under which the costs of open innovation may counteract its benefits. Additionally, much of the research largely ignores the potential role and benefits of closed innovation. In this study, we address these issues by developing a theory related to how the benefits and costs of open innovation lead to an S-shaped relationship between the degree of openness – ranging from closed to low, medium, and high levels of open innovation – and a firm's financial performance. Furthermore, we investigate two possible contingencies in which this relationship is more pronounced: in industries with high appropriability, optimizing firms' ability to extract value from innovation and in dynamic industries, where coordinating high open innovation activities amid rapid changes is exceedingly costly. To test our hypotheses, we create a longitudinal measure for firms' degree of open innovation by using machine-learning content analyses to build an open innovation dictionary and then applying this dictionary to analyze the 10-K annual reports of >9000 publicly listed firms in the U.S. between 1994 and 2017. The results support our theorizing that the relationship between the degree of open innovation and firm financial performance is S-shaped and that industries' appropriability regimes and environmental dynamism are critical boundary conditions for this relationship.