Invasive non-native plants pose a ubiquitous threat to native plant communities and have been blamed for the decline of many endangered species. Endangered species legislation provides legal instruments for protection, but identifying a general method for protecting endangered species by managing non-natives is confounded by multiple factors. We compared non-native management methods aimed at increasing populations of an endangered forb, Ambrosia pumila, and associated native plants. We compared the effects of a grass-specific herbicide (Fusilade II), hand-pulling, and mowing in two degraded coastal sage scrub sites in southern California, U.S.A. At both sites, hand-pulling had the greatest effect on non-native cover, and correspondingly resulted in the greatest increase in A. pumila stems. Fusilade II application also led to an increase in A. pumila, but was not as effective in controlling non-native plants as hand-pulling and its effect varied with the dominant non-native species. Mowing was not effective at promoting A. pumila, and its effect on non-native cover seemed to be related to rainfall patterns. Although some methods increased A. pumila, none of our treatments simultaneously increased cover of other native plants. Hand-pulling, the most effective treatment, is labor intensive and thus only feasible at small spatial scales. At larger scales, managers should take an experimental approach to identifying the most appropriate method because this can vary depending on the specific management objective (endangered species or whole native community), the dominant non-natives, yearly variation in weather, and the timing of treatment application.