Recent national reviews of river restoration have called for rigorous, quantitative evaluations of project success.  Meta-analysis represents one possible tool for testing the efficacy of common restoration practices, as well as determining the role of ecological and methodological variables in explaining inconsistent responses to restoration.  We used meta-analysis to test whether physical habitat restoration increases macroinvertebrate density and richness.  The assumption that macroinvertebrate communities benefit from habitat improvements, sometimes called the ‘field of dreams’ hypothesis, is an underlying tenant of many restoration projects, despite the paucity of quantitative tests of the hypothesis.  A review of the literature identified 43 studies, of which 23 met our selection criteria. Of those, only 9 had sufficient replication necessary to calculate a variance-weighted effect size.  For these 9 studies, adding boulders or large woody debris (LWD) significantly increased richness relative to upstream controls, while density increases were non-significant.  LWD additions had more significant, positive impacts on macroinvertebrate density and richness than boulder additions.  The 14 unreplicated studies also provided evidence for positive mean differences; however differences were not significantly greater than zero.  For studies sampling at multiple points through time, return to pre-disturbance density and richness levels occurred within 100 days post-restoration.