A variety of studies have recognized the incompleteness of occupational health surveillance data in the U.S. Surveillance data are particularly expected to miss certain populations, such as low-wage, contingent, and immigrant workers; and certain health problems, such as diseases with ambiguous symptoms. This study investigated these issues among Cambodian and Lao refugees in Lowell, Massachusetts. Sources of information included selected data from 187 Southeast Asian hospital visits (to Saints Memorial Medical Center, SMMC) charged to workers' compensation; 80 workers' compensation wage replacement cases filed on behalf of Southeast Asian Lowell residents with the Massachusetts Department of Industrial Accidents (DIA) and 160 surveys on self-reports of occupational health collected from 112 Southeast Asian households in Lowell. One hundred two (64%) of survey respondents reported work in manufacturing, and 67 (42%) specifically in electronics assembly. Large proportions of the study population reported exposure to awkward and static postures, repetitive motion, and chemicals such as soldering flux, dusts and solvents. The most common injuries from all three data sources included sprains and strains (37.1% of DIA injuries, 27.4% of SMMC injuries, and 21.9% of survey injuries); cut, laceration, puncture (9.3%, 22.9%, 11.4%); contusion, crushing, bruise (19.6%, 14.6%, 8.7%); and fractures (9.3%, 7.6%, 2.9.%). Symptoms and ill-defined conditions, including overall ill health, headache, dizziness, made up 28.3% of all injuries reported in the survey (95% CI 22.3%,34.3%) but only 1.3% of injuries recorded in the hospital data and none in DIA data. Dermatitis made up 6.8% of survey cases (95% CI 3.5%, 10.1%), but none of the common injuries in the other data sets. Of 155 respondents currently employed, 39 (25%) reported holding a temporary job. Of 147 adults currently employed, 41 (28%) reported regularly working between 41 and 75 hours per week. Twenty-two (14% of respondents) were able to identify sources of information about workplace hazards apart from their supervisors. One hundred ten (69%) of survey respondents expressed no knowledge of workers' compensation. Most sought medical care in small private clinics. Given these employment and medical care patterns, few occupational health problems in this population would be expected in data based on workers' compensation, hospital records, passive surveillance programs, or written questionnaires. Consistent with this expectation, only 4% of workers' compensation cases for Lowell residents filed during the period under study were identified as corresponding to the ethnic groups under study, approximately one eighth the proportion of this group in the Lowell population. Hospital records and workers' compensation records reveal much lower levels of illness and less chemical exposure and awkward/static postures than those reflected in the survey.