The study aims to assess the accuracy of the computerized compensation records for Taiwan's workers with occupational diseases. A total of 233 computerized records registered between 1994 and 1996 were collected and compared with paper records. Among the selected records, 138 were cases of injury or diseases, 53 were disability cases, and 42 were fatal cases. The computerized information on a total of 13 variables was verified against the paper records in the study. The results show inconsistencies between the computerized and the paper records for seven variables including date of development of injury (n = 7), days of work loss (n = 38), amount of money compensated (n = 4), site of injury (n = 2), vector (n = 8), severity of disability (n = 2), and characteristics of disability (n = 2). No inconsistency, on the other hand, was found for personal ID, age, sex, and types of injury or disease. The results also show an obvious conflicting causal link between the content of employment and work-related diseases in 30 records. The findings suggest that the accuracy of Taiwan's computerized compensation records need to be improved, in particular for such variables as site of injury, vector, and date of development of injury which are crucial for the occupational health research. Additionally, we also noted that the information on ICD-code of injury or disease and occupation do not appear in the computerized data, which essentially limits the utility of Taiwan's computerized compensation records from the viewpoint of occupational health research.