Occupational injury is associated with numerous individual and work-related risk factors, including long working hours and short sleep duration; however, the complex mechanisms causing such injuries are not yet fully understood. The authors used structural equation modeling (SEM) as a novel approach to examine fatigue-related direct and indirect potential risk factors for occupational injury. The study sample contained 89,366 employed workers from the National Health Interview Survey (pooled across 6 years, 2004-2009), an annual survey of a representative cross-sectional sample of the US population. Direct and indirect effects of weekly hours worked and usual sleep duration on occupational injuries were modeled using SEM procedures for dichotomous outcomes and a complex sampling design. Confounding and mediating effects of gender, age, race/ethnicity, occupation, industry, type of pay, body mass index (BMI), and psychological distress were simultaneously examined. Long working hours and short sleep duration independently increased the risk of injury. Additional direct risk factors were gender, occupation, type of pay, and BMI. At the same time, sleep duration mediated the adverse relations of long working hours, high psychological distress, and high BMI with injury. These findings indicate that SEM is a useful approach with which to examine dichotomous outcomes and indirect effects in complex samples, and it offers a comprehensive new model of injury prediction.