There has been concern about the rising premiums paid by employers for workers' compensation insurance and possible overuse by workers. It is widely assumed in the business community and by many policy makers that there are a great many people who apply for workers' compensation benefits and do not deserve them but very few who are entitled to benefits but do not apply. We examined a state wide occupational surveillance system that receives reports from individual health care practitioners, hospitals, clinic and employer based health care practitioners. Reporting is required and approximately 20,000 occupational illnesses are received a year. We used both personal interviews and matching of state wide data bases to conduct out study. We found that most workers reported with an occupational disease did not apply for workers' compensation. This was true whether the work-related disease was a chronic disease with long latency (ie. asbestosis or silicosis) or a more acute disease such as repetitive motion injury. Overall, only 25% of workers diagnosed as having a work-related disease applied for workers' compensation. We conducted more in depth analysis of why workers with work-related musculoskeletal disease did not file. Factors associated with filing included increased length of employment, lower annual income, dissatisfaction with co-workers, physical restrictions on activity, type of physician providing treatment, off work ~ 7 days, decreased current health status and increased severity. The significance of our findings are that analyses of the burden of work-related illnesses that rely solely on worker compensation data bases markedly underestimate the true burden of occupational illness. Our results are useful in assessing the true burden of occupational diseases. Programs that evaluate the burden of occupational illnesses and/or the efficacy of preventive interventions will obtain a more accurate and complete picture if they use a more comprehensive surveillance system than a worker compensation claim data base.