In an attempt to restore the connectivity of fragmented river habitats, a variety of passage facilities have been installed at river barriers. Despite the cost of building these structures, there has been no quantitative evaluation of their overall success at restoring fish passage. We reviewed articles from 1960 to 2011, extracted data from 65 papers on fish passage efficiency, size and species of fish, and fishway characteristics to determine the best predictors of fishway efficiency. Because data were scarce for fishes other than salmonids (order Salmoniformes), we combined data for all non‐salmonids for our analysis. On average, downstream passage efficiency was 68.5%, slightly higher than upstream passage efficiency of 41.7%, and neither differed across the geographical regions of study. Salmonids were more successful than non‐salmonids in passing upstream (61.7 vs. 21.1%) and downstream (74.6 vs. 39.6%) through fish passage facilities. Passage efficiency differed significantly between types of fishways; pool and weir, pool and slot and natural fishways had the highest efficiencies, whereas Denil and fish locks/elevators had the lowest. Upstream passage efficiency decreased significantly with fishway slope, but increased with fishway length, and water velocity. An information‐theoretic analysis indicated that the best predictors of fish passage efficiency were order of fish (i.e. salmonids > non‐salmonids), type of fishway and length of fishway. Overall, the low efficiency of passage facilities indicated that most need to be improved to sufficiently mitigate habitat fragmentation for the complete fish community across a range of environmental conditions.