Abstract Restoration practices are often based on trial and error or anecdotal information because data from controlled experiments are not available. In wet meadow restorations of the upper Midwest United States, Reed canary grass (Phalaris arundinacea L.) is controlled with spring burning and spring glyphosate herbicide applications, but the relative effectiveness of either treatment with respect to P. arundinacea growth and life history has not been assessed. We designed a multiyear field experiment to evaluate effects of burning and herbicide application timings on P. arundinacea populations. Burning did not reduce P. arundinacea biomass but reduced the P. arundinacea seed bank, potentially limiting recolonization of P. arundinacea. Glyphosate applications in late August and late September were more effective than in mid‐May (due to enhanced glyphosate translocation to rhizomes), such that two mid‐May applications reduced P. arundinacea biomass to a level equivalent to that achieved by one late‐season application. Phalaris. arundinacea recolonized rapidly from the seed bank and, in plots that received suboptimally timed (mid‐May) herbicide, from rhizomes. Establishment of native species was very low, likely due to competition with recolonizing P. arundinacea. Unplanted species (from the seed bank and refugial populations) accounted for the majority of non‐P. arundinacea biomass. Recolonization of other species was strongly limited by a threshold level of P. arundinacea biomass. Adequate site preparation (over multiple growing seasons) and aftercare (selective removal of P. arundinacea) will be the key to facilitating subsequent wet meadow vegetation establishment. This research provides an example of the importance of experimental evidence as the basis to improve the efficiency of restoration practices.