Back belts have gained popularity under the assumption that their use will reduce low back injuries and thereby decrease the costs. This study sought to examine that hypothesis. A retrospective survey instrument was administered to 1316 workers who perform lifting activities at Tinker Air Force Base, Midwest City, Oklahoma, to identify belt use, lifting requirements, injury, and treatment history. Analysis was also performed on costs applicable to providing the belts, treatment of injury, and lost or limited duty work days. Results show an odds ratio of 1.01 (confidence interval 1.01, 1.02) and P value of .0005 of low back injury with the number and weight of lifts performed in an 8-hour period as well as an odds ratio of 5.56 (confidence interval 3.35, 9.26) with prior history of injury. A protective effect, odds ratio of .65, P value of .019 is noted between lifting and attendance at a back training program. Use of a back belt appears to be marginally effective (odds ratio .60, P value .0508) in reducing injuries when controlling for other related factors. Cost analysis over all diagnoses of low back injury shows less intensive treatment and lower cost per injury for workers injured without a belt versus those injured while wearing a belt. It appears that predictors of low back injury are those expected based on risk assumed (amount of time spent performing lifts, history of injury) and that training programs are effective in prevention of problems. Data suggest that back belts appear to be minimally effective in preventing injury.(ABSTRACT TRUNCATED AT 250 WORDS)