We studied knee-related occupational disability and associated injuries using information recorded in the Total Army Injury and Health Outcomes Database (T AIHOD). Of interest were two different, but related events: hospitalization for severe knee injury and knee-related disability discharge from the Army. The overall objective was to investigate the nature of and differences among the risk factors for these two events. The data source, the T AIHOD, is a relational database containing occupational and selected health information for all individuals on active duty in the U.S. Army, beginning January 1, 1980. The source population for this study consists of all enlisted personnel in the US. Army on active duty at any time between 1980 and 1998, about 2.7 million individuals (11 % women). We analyzed selected sub-samples of the TAIHOD to address three specific aims: 1. What are the risk factors for knee-related disability discharge? 2. What are the risk factors for severe knee injury defined by hospitalization for a kneerelated diagnosis or procedure? 3. How do risk factors for injuries that lead to disability compare to risk factors for injuries that do not lead to disability? For each aim, an appropriate analysis sample of the TAIHOD was constructed using standard methods of incidence density sampling. For aim one, this yielded a sample size numbering 6,810, of whom 1,703 were discharged for knee related disability. For aim two, the analysis sample numbered 12,000, of whom 4,000 were hospitalized for severe knee injury. Aim three entailed a comparison of the results of analyses for the other two aims, with particular attention to ways in which the model for each outcome differed. Analysis results suggested that predictors of knee-related disability discharge from the Army and predictors of knee-related hospitalization were different. For knee-related disability discharge, occupational characteristics were less potent correlates of the outcome than were sociodemographic characteristics. This was especially true for women, for whom no specific work task was associated with odds of knee-related disability discharge. Among both men and women, non-whites had reduced odds of knee-related disability (OR=0.5 for women and OR=0.8 for men), but higher odds of knee injury (OR=1.3 for women, OR=1.2 for men) than whites. For both genders, those in the oldest age quintile had higher odds of both injury and disability compared to those in the youngest age quintile. Even controlling for age and other factors, duration of service was positively associated with odds of knee injury (OR=l.4 for fifth quintile vs. first quintile for each gender). Pay grade was negatively associated with odds of disability (OR=0.2 for E7-E9 vs. EI-E3, men and women) and with odds of injury for men (OR=0.7), in spite of its positive correlation with both age and duration of service. Patterns of odds ratios for other occupational and demographic factors differed for men and women by outcome. Exploratory analyses of possible effect modification suggested interactions between demographic and occupational factors including race/ethnicity, age, and duration of service. We also assessed the adequacy of the final outcome- and gender-specific models for different types of injuries and disabilities. These analyses were possible because of the availability of a large database containing occupational, demographic and health information for a cohort of Army personnel. The wide range of data elements and the large, relatively diverse population enabled the evaluation of the separate and combined influence of sociodemographic and occupational characteristics on the risk of occupational injury and disability. The differences between the final injury and disability models suggests that the use of separate case-control comparisons to identify risk factors for related outcomes is a viable research method.