A prospective epidemiologic study was conducted in two units (140 beds and 57 nursing assistants) of a nursing home to demonstrate the efficacy of an ergonomic intervention strategy to reduce back stress to nursing personnel. The total programme involved the following: determining patient handling tasks perceived to be most stressful by the nursing assistants (NAs); performing an ergonomic evaluation of these tasks; and conducting a laboratory study to select patient transferring devices perceived to produce less physical stress than existing manual patient-handling methods. The intervention phase included training NAs in the use of these devices, modifying toilets and shower rooms, and applying techniques to patient care. Immediately after completing the intervention programme, a post-intervention analysis (which lasted eight months in unit 1 and four months in unit 2) was performed. A biomechanical evaluation of the physical demands required to perform stressful patient-handling tasks showed that the mean compressive force on the L5/S1 disc, the mean hand force required to make a transfer, and the strength requirements (expressed as percentage female population capable) were 1964 N, 122 N, and 83% after intervention as compared to 4751 N, 312 N, and 41% before intervention. Subjectively, the mean rating of perceived exertion was less than 'very light' after intervention as compared to between 'somewhat hard' and 'hard' before intervention. Overall, the mean acceptability rates for the walking belt and the mechanical hoist were 81% and 87% for patient transfers. The incidence rate for back injuries prior to the intervention, 83 per 200,000 work-hours, decreased to 47 per 200,000 work-hours after the intervention. There were no injuries resulting in lost or restricted work days during the last four months of the post-intervention. It is concluded that an appropriate ergonomic intervention programme offers great promise in reducing physical stress and risk of low-back pain to nursing personnel. However, large-scale studies in different nursing homes are needed to confirm the above findings.