1. Despite a global footprint of shifts in flowering phenology in response to climate change, the reproductive consequences of these shifts are poorly understood. Furthermore, it is unknown whether altered flowering times affect plant population viability. 2. We examine whether climate change-induced earlier flowering has consequences for population persistence by incorporating reproductive losses from frost damage (a risk of early flowering) in population models of a subalpine sunflower (Helianthella quinquenervis). Using long-term demographic data for three populations that span the species’ elevation range (8–15 years, depending on the population), we first examine how snowmelt date affects plant vital rates. To verify vital rate responses to snowmelt date experimentally, we manipulate snowmelt date with a snow removal experiment at one population. Finally, we construct stochastic population projection models and Life Table Response Experiments for each population. 3. We find that populations decline (λs < 1) as snowmelt dates become earlier. Frost damage to flower buds, a consequence of climate change-induced earlier flowering, does not contribute strongly to population declines. Instead, we find evidence that negative effects on survival, likely due to increased drought risk during longer growing seasons, drive projected population declines under earlier snowmelt dates. 4. Synthesis. Shifts in flowering phenology are a conspicuous and important aspect of biological responses to climate change, but here we show that the phenology of reproductive events can be unreliable measures of threats to population persistence, even when earlier flowering is associated with substantial reproductive losses. Evidence for shifts in reproductive phenology, along with scarcer evidence that these shifts actually influence reproductive success, are valuable but can paint an incomplete and even misleading picture of plant population responses to climate change.