Population viability analysis (PVA) is commonly used to assess future potential risks to threatened species. These models are typically based on mean vital rates, such as survival and fecundity, with some level of environmental stochasticity. However, the vital rates of wild populations, especially those already exhibiting declining trajectories, may be nonstationary, such that the mean or variance changes over time. In this study, we examined whether including observed temporal trends in vital rates affects the predictive accuracy of PVA, as well as the projected impact associated with a hypothetical threat. To achieve this, we ran a series of simulations using Leslie matrix PVA models that included different combinations of environmental stochasticity, temporal trends in vital rates, and threat. We found that including observed temporal trends in vital rates was (i) crucial for the accurate reconstruction of observed population dynamics and (ii) had a dramatic effect on the projected impact from the hypothetical threat. In an era when many animal and plant populations are declining due to long-term trends in their vital rates, we conclude that this demographic structure is essential for robustly evaluating potential threats using PVA models. Omitting observed temporal trends in vital rates from impact assessments is highly likely to yield unreliable results that could misinform conservation and management decision making.