The aim of this study is to determine the validity of the ICD-9 for diagnostic classification of soft tissue disorders of the neck and upper limb, using routinely collected data at a large steel company. The documentation in the clinical chart served as the gold standard. First, the overall accuracy of identifying these disorders from the ICD-9 was examined. Second, we examined whether the codes themselves, on an individual basis, accurately reflected the underlying problems as documented in the medical records. There were 1,267 new cases identified in 1991 by inclusion of all potentially applicable ICD-9 codes. Only 805 (63.5%) fulfilled the definition of a soft tissue disorder of the neck or upper limb as determined by chart review. A more restrictive strategy that only included cases coded by those ICD-9 codes that specifically pertain to these disorders yielded a higher proportion of true cases 458/480 cases (95.4%), but failed to identify the other 347 cases. The anatomical site of the problem could not be identified from the codes describing 651/1,267 cases (51.4%). There was poor agreement between the diagnostic labels recorded in the medical records and the ICD-9 codes, suggesting that many of the terms are being used interchangeably. Our results suggest that conclusions about these disorders drawn from analysis of administrative data which rely upon the ICD-9 for diagnostic classification must be interpreted cautiously. For these soft tissue disorders, researchers will need to develop strategies which would improve upon and supplement the ICD-9.