Union administrative records were combined with workers' compensation data to identify a cohort of 12,958 active union carpenters, their person-time at risk, and their documented work-related eye injuries between 1989 and 1995 in the state of Washington. The injuries were described using ANSI codes for injury nature, type (mechanism), and source or object associated with the event. Injuries which resulted in paid lost time from work were also described based on the ICD-9 codes attached to claims for their medical treatment. Overall rates of filing compensation claims for eye injuries as well as age, gender, and union local specific rates were calculated. To identify high risk subgroups and explore incident and recurrent events, the person-time and events were stratified by age, gender, time in the union, claim status, and predominant type of work of the union local with which each carpenter was affiliated for multivariate analyses with Poisson regression. Eye injuries were responsible for 12 percent (n = 1730) of workers' compensation claims during this time period, exceeded only by back and finger injuries. Thirty-one claims resulted in paid lost time from work and these cases accounted for one-third of all costs for medical care for eye injuries. At least 10 percent of all medical costs for eye injuries and 35.5 percent of medical costs for eye injuries which resulted in paid lost time were associated with injuries sustained while hammering--a very common carpenter exposure. Claims were filed at an estimated rate of 6.1 per 200,000 hours worked. Individuals with previous compensation claims for eye injuries had rates of injury 1.6 times higher than individuals without previous eye injuries. Rates decreased significantly with age and time in the union. Eye injuries among these union carpenters were very common, but the rate of injuries severe enough to require paid time off work was quite low. These findings raise questions about factors which might influence the failure to use appropriate protection including availability and acceptability of eye protection, use by peers, and perception of risk.