Although many people have expressed alarm that we are witnessing a mass extinction, few projections have been quantified, owing to limited availability of time-series data on threatened organisms, especially plants. To quantify the risk of extinction, we need to monitor changes in population size over time for as many species as possible. Here, we present the world's first quantitative projection of plant species loss at a national level, with stochastic simulations based on the results of population censuses of 1618 threatened plant taxa in 3574 map cells of ca. 100 km2. More than 500 lay botanists helped monitor those taxa in 1994-1995 and in 2003-2004. We projected that between 370 and 561 vascular plant taxa will go extinct in Japan during the next century if past trends of population decline continue. This extinction rate is approximately two to three times the global rate. Using time-series data, we show that existing national protected areas (PAs) covering ca. 7% of Japan will not adequately prevent population declines: even core PAs can protect at best <60% of local populations from decline. Thus, the Aichi Biodiversity Target to expand PAs to 17% of land (and inland water) areas, as committed to by many national governments, is not enough: only 29.2% of currently threatened species will become non-threatened under the assumption that probability of protection success by PAs is 0.5, which our assessment shows is realistic. In countries where volunteers can be organized to monitor threatened taxa, censuses using our method should be able to quantify how fast we are losing species and to assess how effective current conservation measures such as PAs are in preventing species extinction.