Demographic models are an invaluable tool for understanding and describing population dynamics, but they are also widely used in conservation for estimating extinction risks and predicting management needs. In these types of studies, however, the amount of data available is often limited, leading to the difficult question of how much data is needed to provide accurate estimates of demographic parameters. Using a unique 31-year dataset for a population of *Frasera speciosa* (Gentianaceae) in the Elk Mountains of southwestern Colorado, we built stage-based population matrix models in order to: 1) describe the demography of this long-lived monocarpic plant, 2) compare the results from using a deterministic and a stochastic modeling approach, and 3) examine how the length of the dataset used in the demographic model can affect the accuracy of estimating population parameters.**Results/Conclusions **

Using the entire long-term dataset in a deterministic model, we find that the *F. speciosa *population is likely stable, with a population growth rate (or lambda) of 1.002. The estimated generation time is 51.52 years, and the transitions representing stasis and growth contribute the most to the rate of population growth (their summed elasticity values are 0.57 and 0.29, respectively). In agreement with a previous study, the lambda estimates from the deterministic models tend to be higher than those from the stochastic models. When comparing the accuracy of the parameter estimates between models built using different lengths of the dataset, we find that lambda estimates based on more than 20 years of data deviate less than 1% from the long-term estimate (based on the entire 31-year dataset).