Because tornadoes cause enormous damage to humans and societies, projecting plausible future changes in tornado activity is an important research focus. We used the results of climate experiments under the A1B emissions scenario with a 20-km-mesh and a high-resolution atmospheric global circulation model to project future changes in the frequency of conditions conducive to the generation of strong tornadoes (F2 or greater on the Fujita scale) in Japan. We found that this frequency is likely to double in future in almost all areas of the Japanese Islands in spring and on the Japan Sea side of the Japanese Islands in summer due to intensification of atmospheric instability caused by an increase in the water-vapor mixing ratio and a temperature rise in the lower troposphere. In contrast, the frequency of strong vertical wind shear, which is conducive to tornadogenesis, was projected to hardly change or decrease slightly. Comparison with climate fields generated by a 60-km-mesh 12-member ensemble experiment showed that future changes in tornadogenesis-favorable environmental conditions projected by the 20-km-mesh experiment were highly reliable. Moreover, we found that the predicted future changes were robust when we used other thresholds for the environmental parameters that created conditions conducive to strong tornadoes. Our results indicate that there will be a significant increase in the frequency of strong tornadoes in Japan in the future.