We re-evaluated a method to estimate water stress changes as a function of increasing global mean temperature. Trends in the total population under high water stress (defined as when the annual water withdrawal divided by the annual water availability is higher than 0.4) now and in the future (total HWSP) and the population exposed to high water stress in the future but not now (add_HWSP) are dependent on differences in each scenario, not the global mean temperature increase. The ensemble mean of the total HWSP and its ratio for emissions scenarios A1B and B1 are close to constant values and decreased when the global mean temperature increase exceeded 1.9°C and 1.3°C, respectively. If the global mean temperature increase reaches a maximum ensemble-mean value in each scenario, the total HWSP (its ratio) in scenarios A1B, A2 and B1 was about 4.5 billion (0.54), 8.6 billion (0.64) and 4.1 billion (0.49), respectively. We estimated the contributions of climate change to runoff, water withdrawal and population growth on total HWSP and add_HWSP to separate the influences of climate change and socio-economic change. Climate change and socio-economic factors (water withdrawal and population growth) decreased and increased add_HWSP, respectively.