The categorization method was used to evaluate the performance of the Coupled Model Inter-comparison Program Phase 5 (CMIP5) climate models in simulating the decadal variation of summer monsoon rainfall in eastern China. Out of the 38 models that were examined, only six (category-1 models) captured the decadal variation of summer rainfall in eastern China in the late 1970s, i.e., increased rainfall in the Yangtze River basin and decreased rainfall in North and South China. The category-1 models simulate the decadal variation of summer rainfall in eastern China well because they are able to capture the decadal weakening of the East Asian Summer Monsoon (EASM) along with the northerly anomaly in the coast of East Asia and changes in the western Pacific subtropical high in the late 1970s. In contrast, the category-2 models, which poorly simulate the decadal variation of summer rainfall in eastern China, fail to reproduce the decadal weakening of the EASM. Further analysis indicates that while most models can reproduce the spatial distribution of the Pacific Decadal Oscillation (PDO), their accuracy at reproducing the decadal variation of the PDO varies. Category-1 models can reproduce the tropical ocean warming in the late 1970s and the relative phase transition of the PDO. Category-2 models fail to capture the phase transition of the PDO, the tropical ocean warming in the central-eastern Pacific Ocean, and the cooling trend in the Yangtze River basin, which results in low accuracy of these models for reproducing the weakened EASM and the decadal variation of summer rainfall in eastern China after the late 1970s. This indicates that the accuracy of climate models for simulating the decadal variations of sea surface temperature plays an important role in the accuracy of the models' simulation of the decadal variation of summer rainfall.