Projections of the magnitude and pattern of possible health risks from climate change should be based on multiple climate and development scenarios to describe the range of uncertainties, to inform effective and efficient policies. For a better understanding of climate change-related risks in seven metropolitan cities of South Korea, we estimated temperature-related summer (June to August) mortality until 2100 using projected changes in climate, population, and adaptation. In addition, we extracted the variations in the mortality estimates associated with uncertainties in climate, population, and adaptation scenarios using 25 climate models, two Representative Concentration Pathways (RCP 4.5 and 8.5), three population scenarios (high, medium and low variants), and four adaptation scenarios (absolute threshold shift, slope reduction in the temperature-mortality relationship, a combination of slope reduction and threshold shift, and a sigmoid function based on the historical trend). Compared to the baseline period (1991–2015), temperature-attributable mortality in South Korea during summer in the 2090s is projected to increase 5.1 times for RCP 4.5 and 12.9 times for RCP 8.5 due to climate and population changes. Estimated future mortality varies by up to +44%/−55%, −80%, −60%, and +12%/−11% associated with the choice of climate models, adaptation, climate, and population scenarios, respectively, compared to the mortality estimated for the median of the climate models, no adaptation, RCP 8.5, and medium population variant. Health system choices about adaptation are the most important determinants of future mortality after climate projections. The range of possible future mortality underscores the importance of flexible, iterative risk management.