Greenhouse-gas emissions are indirectly causing future deaths by multiple mechanisms. For example, reduced food and water supplies will exacerbate hunger, disease, violence, and migration. How will anthropogenic global warming (AGW) affect global mortality due to poverty around and beyond 2100? Roughly, how much burned fossil carbon corresponds to one future death? What are the psychological, medical, political, and economic implications? Predicted death tolls are crucial for policy formulation, but uncertainty increases with temporal distance from the present and estimates may be biased. Order-of-magnitude estimates should refer to literature from diverse relevant disciplines. The carbon budget for 2°C AGW (roughly 10¹² tonnes carbon) will indirectly cause roughly 10⁹ future premature deaths (10% of projected maximum global population), spread over one to two centuries. This zeroth-order prediction is relative and in addition to existing preventable death rates. It lies between likely best- and worst-case scenarios of roughly 3 × 10⁸ and 3 × 10⁹, corresponding to plus/minus one standard deviation on a logarithmic scale in a Gaussian probability distribution. It implies that one future premature death is caused every time roughly 1,000 (300-3,000) tonnes of carbon are burned. Therefore, any fossil-fuel project that burns millions of tons of carbon is probably indirectly killing thousands of future people. The prediction may be considered <i>valid</i>, accounting for multiple indirect links between AGW and death rates in a top-down approach, but <i>unreliable</i> due to the uncertainty of climate change feedback and interactions between physical, biological, social, and political climate impacts (e.g., ecological cascade effects and co-extinction). Given universal agreement on the value of human lives, a death toll of this unprecedented magnitude must be avoided at all costs. As a clear political message, the "1,000-tonne r Nolt (2011a) found that the emissions of a typical US-American over the course of a lifetime cause the suffering or death of one or two future people in poor countries during the next millennium. Of course, not only Americans are to blame, although their emissions are roughly twice those of Europeans; in round figures, 10 9 rich people are in the process of prematurely ending the lives of 10 9 future poor people. Nolt (2015) summarized other attempts to estimate future death tolls in connection with AGW, concluding that “the methodologies of these studies are imperfect … clearly there is need for critique and further refinement of such estimates” (p. 353). Sekerci and Petrovskii (2015) showed that “the oxygen production by marine phytoplankton can stop suddenly if the water temperature exceeds a certain critical value. Since the ocean plankton produces altogether more than one half of the total atmospheric oxygen, it would mean oxygen depletion not only in the water but also in the air. Should it happen, it would obviously kill most of life on Earth” (p. 2349). Soil will be degraded by chemical-heavy farming techniques and deforestation-induced erosion, reducing crop yields ( Arsenault, 2014 ). Given the multiple serious effects of AGW on global food and water supplies, and its expected effect on international conflict and migration, the death rate attributable to AGW (the difference between the two lines) may approach 10 7 per year by 2100. If so, the total AGW death toll in the 21st century will be 5 × 10 8 (the area between the two lines). Assuming another 5 × 10 8 climate deaths in the 22nd and later centuries, the total death toll due to historic and current emissions will be roughly 10 9 . The 1,000-Tonne Rule If burning 10 12 tonnes of fossil carbon causes the premature deaths of 10 9 future people, one future person is killed every time a 1,000 tonnes of carbon are burned. This OME is consistent with Nolt’s (2011a) finding that the average American produces 1,840 tonnes CO 2 equivalent (from burning 500 tonnes of carbon) during her or his lifetime, which then causes the suffering or death of one or two future people. If it takes 1,000 tonnes of carbon (or 3,700 tonnes of CO 2 ) to kill a future person, and Nolt’s calculation is correct, the average American is killing half of a future person. The 1,000-tonne rule can be understood by the following thought experiment. Imagine what would happen if humanity suddenly and completely stopped producing GHGs. The 1,000-tonne rule applies to the left portion of the graph, before the first 10 12 tonnes of carbon have been burned. Consider a medium-level “business as usual” scenario (e.g., IPCC Assessment Report 5, Working Group II, Representative Concentration Pathways) with a GMST increase of 3–4°C by 2100. If business as usual were continued into the future until all reasonably accessible carbon reserves (5 × 10 12 tonnes according to Tokarska et al., 2016 ) were exhausted, GMST would increase by some 10°C, killing most or all humans ( Sherwood and Huber, 2010 ; the right side of Figure 3 ). In that extreme case, roughly 500 tons of carbon will have been sufficient to kill one person. Conclusion Convergent evidence from diverse approaches suggests that 2°C AGW, caused by burning 10 12 tonnes of carbon (equivalent), will kill roughly 10 9 people over a period of one to two centuries. Therefore, one future person is killed every time roughly 10 3 tonnes of carbon are burned. This “1,000-tonne rule” implies that every fossil fuel industry whose productivity can be expressed in millions of tonnes of burned carbon is causing the premature deaths of thousands of future people. Consider the following examples: A typical passenger jet carries 300,000 l of fuel and consumes 200,000 on a long flight, creating 500 tonnes of CO 2 corresponding to 135 tonnes of carbon. Cigarettes currently kill 7 × 10 6 people per year ( Rentería et al., 2016 ). Comprehensive cigarette advertising