Tropical spiderwort (Commelina benghalensis L.) is a noxious invasive species and was detected in a long-term experiment in a research farm in Goldsboro, NC. A multistakeholder governance model was used to address the invasion of this species. Regulators insisted on the use of fumigation in all fields, but after intense negotiations, a multi-tier eradication plan was designed and implemented, allowing fumigation outside the long-term experiment and a combination of integrated approaches (including physical removal) and intense monitoring and mapping for long-term experimental fields. In the long-term experiment, C. benghalensis populations decreased logarithmically from more than 50,000 plants in approximately 80 ha in 2005 to 19 plants in less than 1 ha in 2019, with a projection of full eradication by 2024. Despite these results, which were considered to be proof of successful ecological management by university researchers, regulators decided to fumigate the fields containing the remaining 19 plants. This decision was made because regulators considered factors such as professional liability and control efficacy. This created serious disagreements between the different stakeholders who participated in the design of the original plan. Despite the goodwill all parties exhibited at the beginning of the governance process, there were important shortcomings that likely contributed to the disagreements at the end. For example, the plan did not include specific milestones, and there was no clarity about what acceptable progress was based on (i.e., plant numbers or the rate of population decline). Also, no financial limits were established, which made administrators concerned about the financial burden the eradication program had become over time. Multistakeholder governance can effectively address plant invasions, but proper definition of progress and the point at which the program must be modified are critical for success, and all this must be done within a governance model that balances power in the decision-making process.