Methods for using simple population count data to project extinction risk have been the focus of much recent theoretical work, but few researchers have used these approaches to address management questions. We analyzed 15 years of census data on the federally endangered endemic riparian plant Pityopsis ruthii (Small) with the diffusion approximation (DA). Our goals were to evaluate relative extinction risk among populations in two different watersheds (in Tennessee, U.S.A.) and potential effects of variation in managed river flow on population dynamics. Populations in both watersheds had high projected risks of extinction within 50 years, but the causes of this risk differed. Populations of P ruthii on the Hiwassee River had higher initial population sizes but significantly lower average growth rates than those on the Ocoee River The only populations with low predicted short-term extinction risk were on the Ocoee. Growth rates for populations on both rivers were significantly reduced during periods of lower river flow. We found only marginal evidence of a quadratic relationship between population performance and flow. These patterns are consistent with the idea that low flows affect P ruthii due to growth of competing vegetation, but the degree to which very high flows may reduce population growth is still unclear Simulations indicated that populations were most sensitive to growth rates in low-flow years, but small changes in the frequency of these periods did not strongly increase risk for most populations. Consistent with results of other studies, DA estimates of extinction risk had wide confidence limits. Still, our results yielded several valuable insights, including the need for greater monitoring of populations on the Hiwassee and the importance of low-flow years to population growth. Our work illustrates the potential value of simple methods for analyzing count data despite the challenges posed by uncertainty in estimates of extinction risk.