Occupational injury rates in the US have been in steady decline through the 1990s. After an increase in the injury rate from 1983 to 1992, the rate of non-fatal injures in the US workplace declined from 8.3 cases per 100 full-time employees in 1990 to 5.9 cases in 1999, a 29% drop. The composition of occupational injuries also appears to have changed in the '90s. Specifically, the rate of cases involving only restricted work has risen by 50% (from 0.8 in 1992 to 1.2 in 1999) while those more serious injuries involving days away from work have declined by almost 30% (from 3 in 1992 to 1.9 in 1999). Our research probed this trend, its causes, and its economic consequences. As proposed, we merged data from the Bureau of Labor Statistics' (BLS) Annual Surveys of Occupational Injury and Illness with other establishment-level data on training, benefits, and anti-drug programs to analyze the effect of a wide range of policies and establishment characteristics on the reported injury rate. We find that the reasonable accommodations mandate of the Americans with Disabilities Act (ADA) is linked to an increase in the rate of restricted work during the 1990s. The rate of cases with no lost work is also associated with the accommodations mandate suggesting a possible mis-reporting of restricted work injures as no-lost-work cases. Unexpectedly, the ADA's accommodation provision also appears to increase the rate of days away from work cases though this effect appears to be limited to large firms. Besides ADA effects, workers' compensation changes in the 1990s that have restricted the compensability of injuries appear to uniformly depress the reported rate of all types of occupational injuries. Our research on the relationship between safety training and occupational injuries suggest that safety training increases the reporting of injuries but also has real safety effects on days-away injuries, especially in smaller firms. Safety training appears to be more effective in preventing severe injuries in large firms than in small ones. We find a 10% lower rate of injuries involving lost-work in the full sample of establishments that used drug testing. Firms with drug testing have a 17% lower rate of minor injuries involving no lost work with the biggest difference (43%) occurring in the transportation and public utility industry. Controlling for other characteristics, the presence of an Employee Assistance Program (EAP) was associated with a significant reduction in no lost-work injuries across firms of different sizes and in the manufacturing industry while being only weakly associated with more serious lost-work cases. However, lost work injuries appear to be more responsive to specific EAP characteristics. Finally, we applied an input-output model to estimate the effect of the reduction in the occupational injury rate on the national employment and the gross domestic product (GDP). We estimate that declining occupational injury increased employment in 2002 by 550,000 jobs. Without the decline in occupational injuries, the unemployment rate in 2002 would have been 6.16% instead of 5.78%. The increase in GDP was $25.5 billion or 9% of the average annual GDP increase from 1993 to 2002.