Double cantilever beam (DCB) tests were conducted by immersing the specimens in temperature-controlled water while applying a creep load using a spring. By introducing a data reduction scheme to the spring-loaded DCB test method, it was confirmed that only a single parameter measurement was sufficient to calculate the energy release rate (ERR). Aluminum alloy substrates bonded with an epoxy adhesive were used, and DCB tests were performed by changing the initial load values, spring constants, and immersion temperatures for two types of surface treatment. The initial applied load and spring constant had no effect on the ERR threshold. In contrast, the threshold decreased with the increasing immersion temperature, but even in the worst case, it was 15% of the critical ERR in the static tests. Using the creep crack growth relationship, it was revealed that there were three phases of creep immersion crack growth in the adhesive joints, and each phase was affected by the temperature. The spring-loaded DCB test method has great potential for investigating the combined effects of creep, moisture, and temperature, and this study has demonstrated the validity of the test method. The long-term durability of adhesive joints becomes increasingly important, and this test method is expected to become widespread.