Owing to declines in salmonid populations, in-stream restoration structures have been used for over 80 years to increase abundance of fish. However, the relative effectiveness of these structures remains unclear for some species or regions, partly due to contrasting conclusions from two previous meta-analyses. To update and reconcile these previous analyses, we conducted a meta-analysis using data available from 1969 to 2019 to estimate the effect of in-stream structures on salmonid abundance (number and density) and biomass. Data from 100 stream restoration projects showed a significant increase in salmonid abundance (effect size 0.636) and biomass (0.621), consistent with previous reviews and studies, and a stronger effect was found in adults than in juvenile fish. Despite a shift towards using more natural structures (wood and boulders) since the 1990s, structures have not become more effective. However, most projects monitor for less than 5 years, which may be insufficient time in some systems for channel morphology to adjust and population changes to be apparent. Process-based techniques, which give more space for the river, allow more long-term, self-sustaining restoration.