We evaluated the reliability of CTC v 2.0 based on source documents and also studied the degree of inconsistency in toxicity grading. Five clinical research coordinators from the National Cancer Center Hospital independently reviewed source documents from 17 patients and graded toxicities in the following common adverse events: diarrhea, nausea, stomatitis/pharyngitis, vomiting, febrile neutropenia, infection, infection unknown source, and sensory neuropathy. If grading was already documented on the medical chart, it was masked so that the coordinator could perform the evaluation without information bias. After the completion of toxicity grading, the participating coordinators discussed each case, and a consensus was reached for final toxicity grading. The proportion of agreement for each toxicity criteria are as follows: diarrhea; 0.59 (95%CI 0.35-0.82), nausea; 0.47 (0.23-0.71), stomatitis/pharyngitis; 0.59 (0.35-0.82), vomiting; 0.71 (0.49-0.92), febrile neutropenia; 0.88 (0.73-1.04), infection; 0.82 (0.64-1.01), infection by unknown source; 0.82 (0.64-1.01), sensory neuropathy; 0.65 0.42-0.87). The cause of variability largely depended on the differences in individual clinical assessment, and misunderstanding of toxicity criteria by coordinators has been observed. Even in a single institution environment, variability exists in the toxicity assessment and grading. Good training and education on toxicity assessment using common criteria and development of translated manual, including the interpretation of criteria assessment, may help reduce variability.