In recent decades, dozens of studies have involved attempts to introduce native and desirable nonnative plant species into grasslands dominated by invasive weeds. The newly introduced plants have proved capable of establishing, but because they are rarely monitored for more than four years, it is unknown if they have a high likelihood of persisting and suppressing invaders for the long term. Beyond invaded grasslands, this lack of long-term monitoring is a general problem plaguing efforts to reintroduce a range of taxa into a range of ecosystems. We introduced species from seed and then periodically measured plant abundances for nine years at one site and 15 years at a second site. To our knowledge, our 15-year data are the longest to date from a seeding experiment in invaded, never-cultivated grassland. At one site, three seeded grasses maintained high densities for three or more years, but then all or nearly all individuals died. At the second site, one grass performed similarly, but two other grasses proliferated and at least one greatly suppressed the dominant invader (Centaurea maculosa). In one study, our point estimate suggests that the seeded grass Thinopyrum intermedium reduced C. maculosa biomass by 93% 15 years after seeding. In some cases, data from three and fewer years after seeding falsely suggested that seeded species were capable of persisting within the invaded grassland. In other cases, data from as late as nine years after seeding falsely suggested seeded populations would not become large enough to suppress the invader. These results show that seeded species sometimes persist and suppress invaders for long periods, but short-term data cannot predict if, when, or where this will occur. Because short-term data are not predictive of long-term seeded species performances, additional long-term data are needed to identify effective practices, traits, and species for revegetating invaded grasslands.