In the past three decades, the role of matrix-based demographic models in plant conservation has steadily increased. However, the reliability of these methods remains hotly debated. Most tests of model performance have relied on strict conditions for either the datasets being tested or the criteria used to judge accuracy of the results. This leads to a potential disconnect between the variety of ways in which models are used in practice and the limited set of conditions where their performance has been evaluated. Our working group brings together a group of ecologists who have worked with these models in applied settings. We will review how models have actually been used in the recent past and discuss what predictions we expect these models to usefully provide. We will then use our demographic data from long-term studies to evaluate how well demographic models actually predict the dynamics of perennial plant populations. We will also address whether increasing methodological complexity (e.g. density dependence, integral projection modeling) improves reliability. The convergence of our group occurs at a moment when sufficient time and data have accumulated to test the predictions of demographic models at relevant time scales for management, and takes advantage of NCEAS capacities to bring together diverse groups and archive key data. Thus, this working group provides a timely opportunity to reevaluate what has become an exceptionally important tool in conservation and management. collapse