Abstract Based on U.S. data for the 48-year-period 1953–2000, this study makes a contribution on the R&D-growth relation along five dimensions. First, we note several descriptive patterns that may be regarded as stylized facts relative to R&D outlays in the U.S. during the half-century period. These include (a) a dramatic increase in the share of non-federal R&D outlays, (b) a corresponding decline in the share of federally funded R&D expenditure, and (c) an even more dramatic decline in the share of defense R&D spending. Second, in a departure from most of the literature on the topic, we study the R&D-growth nexus at a disaggregated level by considering the roles of federal, non-federal, and defense R&D outlays. Third, we use the relatively new bounds-testing and ARDL (autoregressive distributed lag) procedures of Pesaran et al. [Pesaran, M. H., Shin, Y., & Smith, R. J. (2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16, 289–326] to estimate the (long run) relation between R&D outlays and growth in a fairly standard model. Fourth, contrary to the almost universal belief, our estimates indicate a larger role of federal R&D relative to non-federal R&D in growth, and also a stronger role of defense R&D than of non-defense (federal) R&D. Last, to the extent our estimates are reasonable, the above-noted temporal movements in the shares of federal, non-federal, and defense R&D outlays seem to reflect socially perverse trends in the context of economic growth and well-being, and indicate the need for appropriate policy interventions for a substantial enhancement of federal defense R&D and non-defense R&D outlays.