Information on the etiology of work-related injuries among construction workers has been limited by practical problems which make the study of their health and safety hazards difficult; these are especially salient when considering those who do residential and drywall work. These are very mobile workforces with individuals frequently changing job sites and even employers. In residential construction, particularly, the duration of work at any given site is shorter in nature than in commercial construction and the nature of the work changes from day to day. Job sites are typically smaller with fewer workers at any given site. We report on an active injury surveillance project designed to test the utility and feasibility of active injury investigations in identifying causes of work-related injury among a large cohort of residential and drywall carpenters. The program was designed to document the magnitude of injuries among these carpenters, to describe in detail the nature of their injuries and the circumstances surrounding these events. In addition, to explore risk factors for prolonged loss of time from work after a back injury among these high risk construction workers, comparisons were made between injuries, and workers experiencing these injuries, which resulted in rapid return to work and those which resulted in more prolonged work absence (> 1 month). A group of 20 contractors were recruited to participate in the surveillance program, agreeing to report OSHA recordable injuries to the project office as they occurred on their work sites. Experienced journeymen carpenters with safety training and training in questionnaire administration and informed consent interviewed the injured carpenters about the circumstances surrounding their injuries. These carpenter investigators also conducted site assessments, using a standard format, where falls occurred. The union provided enumeration of the carpenters working for these contractors, their union status (apprentice vs journeyman) and hours worked by person by contractor by month, providing person-hours of work as a measure of time at risk; this allowed the estimation of injury rates. These data were supplemented with a series of focus groups designed to collect information about exposures of apprentices, training and mentoring in skills and safety training, perception of risk, and job stressors. In addition, a small group of self-insured contractors provided their workers' compensation data for the years 1995-2000 for analyses. The latter provided some information on costs associated with work related injuries among residential carpenters. After a period of pre-testing, active surveillance data were collected over 37 months beginning in September of 1999. The dynamic cohort consisted of5,137 carpenters who worked for one of20 participating contractors during this time period, representing a total of9,346,603 carpenter hours. Between September 1, 1999 and September 30, 2002 a total of 783 injuries were reported. Of these, 586 injured carpenters participated in injury investigation interviews ( 74.8%). Injuries were most commonly caused by the carpenter being struck by or against something, manual materials hand1ing tasks or some other type of overexertion, falls from elevations, and falls from the same level. Injuries involved the upper extremities over 40% of the time followed by the lower extremity, axial skeleton/trunk, and the head and face including the eyes. The injuries that resulted from being struck by or against something were largely cuts, puncture wounds, scratches/abrasions (including eye injuries), and contusions (87%). Nearly 90% of the overexertion injuries resulted in sprains and strains. Falls from height and same level falls most often resulted in sprains/strains or contusions, but 26% and 17% of these falls, respectively, resulted in fractures. The most common struck by injuries involved pneumatic nail guns. Materials being handled at the time of overexertion injuries were most commonly associated with handling of building materials. The carpenter was handling an object weighing greater than 100 pounds 48% of the time and 200 or more pounds 28% of the time. Carpenters fell from a wide variety of surfaces most commonly ladders, scaffolds and unsecured work surfaces. While some falls were related to challenges in residential building with lack of appropriate anchor points, the vast majority could have been prevented through use of recognized fall prevention and protection strategies such as use of guardrails, covering openings, and appropriate ladder and scaffold use. Same level falls were often related to weather, housekeeping or terrain issues, such as tripping over debris, difficult work terrain (rocky, muddy, uneven); the slope of lot, lack of backfill around the foundation, difficult access, and/or egress from the building. The estimated overall injury rate was 16.8 per 200,000 hours worked (783/9.3 million hours). There were 290 injuries that resulted in lost time from work beyond the day of injury (50% of those interviewed; data item missing for 37), representing a lost-time injury rate of 6.3 per 200,000 hours worked based only on individuals who were interviewed. If those who participated were representative of the pool of injuries, this rate would be as high as 8.4 per 200,000 hours. The injury rate among apprentices was 16.3 per 200,000 hours worked ( 95% CI 14.3 to 18.4) compared to 10.8 per 200,000 hours worked (95% CI 9.7 to 12.1) among journeymen (RR=1.5). Rates were significantly higher overall among apprentices and for injuries that resulted from being struck by or against something (RR=1.9). Nail gun injuries, in particular, occurred at rates 3.1 times higher among apprentices. A significant proportion of back injuries, nearly 30%, were the result of acute trauma from falls or being struck. Among the manual materials handling back injuries there was some indication that the injuries resulting in prolonged loss of time from work were associated with inciting events that created greater acute spinal loads. Data from the self-insured homebuilders group for a six year period were consistent with the active surveillance reports. These compensation data demonstrated falls from elevations to be the most costly injuries; they were responsible for the greatest overall costs (even though they ranked third in frequency) and the greatest cost per claim. Fall rates declined 46% between 1997 and 2000, and total" costs for falls fell to a rankof4th in 1998, 2nd in 1999, and again 4th in 2000. Mean costs per fall were markedly down in 2000 averaging about $7500 per fall compared to a range in previous years of $30,000 per fall (1995) to $12,000 per fall (1998). Overexertion, largely involving manual materials handling, injuries were up 20% in 2000 compared to 1995. Also consistent with the active surveillance findings at least 26% of costs for back injuries that were not the result of manual materials handling tasks or exertion. These injuries were the result of falls, being struck by walls, or slipping. The greatest costs for back injuries were from injuries resulting from lifting framed walls or setting steel beams. Injured carpenters insights into what contributed to their injuries varied by type of injury, but time pressures and speed of work were the most common factors carpenters acknowledged. Forty seven percent (47%) of workers who experienced a same level fall felt time pressures contributed to the circumstances leading to injury, and 20% attributed housekeeping issues. Overexertion injuries were most often attributed to the task being too heavy (34%) or the carpenter needing help (25%). Surveillance, such as this, yields information on factors that contribute to injuries among high risk construction workers who are difficult to study for practical reasons. The active surveillance is more time consuming than passive surveillance activities, but the information is more useful for understanding the circumstances surrounding injuries and in formulation of concrete preventive recommendations. There are challenges in identifying methods to capture injuries completely and in a timely manner to allow rapid investigation.