Curtis Prairie is an 80-yr-old restored prairie that, despite its diverse vegetation and dense cover, was invaded by Phalaris arundinacea (reed canarygrass) during the 1980s. Our search for a method to eradicate Phalaris involved field tests of a grass-specific herbicide (sethoxydim as Vantage®). To reduce Phalaris growth and favor native vegetation, we used an adaptive approach that began in 2005 by testing sethoxydim application with and without adding a seed mix of 32 native plants. Sethoxydim reduced flowering but not cover of Phalaris; we therefore reapplied sethoxydim to all the 2005 plots in 2006 and again in 2007. We began a second experiment in 2006 that combined late-May burning followed by late-June sethoxydim application. This combination reduced Phalaris cover the most of all treatments. In a third experiment in 2007 we repeated the late-May burn + late-June sethoxydim treatment but did not find a reduction in Phalaris cover. We attribute differences in burning + herbicide effects to year effects. Because the Phalaris canopy remained dense even where we applied sethoxydim for 3 yr (first experiment, 2005), we were not surprised that our 2005 seeding of 32 native forbs and graminoids (but not grasses) barely increased native species presence. Six years after seeding, however, we found 12 of the seeded species along with 21 unseeded natives. The adaptive approach of designing new experiments based on prior results led to a promising approach in 2006 (burning before applying sethoxydim to the resprouting Phalaris). Repeating that combination in 2007 indicated a year effect and that no single approach would eradicate this clonal grass.