Engineered in-stream structures are often installed to increase salmonid abundance, either for commercial gain in fisheries or for conservation purposes in degraded habitats. Having been in widespread use for the last 80 years, at an estimated cost of hundreds of millions of U.S. dollars each year, the effectiveness of these structures is still widely debated in the literature. Many studies of varying quality have been undertaken that attempt to address this issue, but it has proved difficult for practitioners to develop a consensus regarding the utility of these structures, despite their continued use. Systematic review methodology was used to formally synthesize empirical evidence regarding the effectiveness of engineered in-stream structures as a management tool to increase salmonid abundance. Meta-analysis shows that evidence regarding the effectiveness of in-stream devices is equivocal. Heterogeneity is significant both for population size and local habitat preference. This heterogeneity is related to stream width, with in-stream devices being less effective in larger streams. Consequently, widespread use of in-stream structures for restoration, particularly in larger streams, is not supported by the current scientific evidence base.