The purpose of this study was to evaluate the diagnostic equivalence, radiation dose, clinical usefulness and radiographic aspects of a low-dose, full-body digital X-ray machine in a busy trauma unit. A digital trauma X-ray machine known as "LODOX" was compared with conventional radiography between June 1999 and November 2001 in the Groote Schuur Hospital Trauma Unit, Cape Town. Digital images of a variety of body regions commonly imaged in trauma were compared for diagnostic image quality in a number of categories with equivalent conventional radiographs. A seven-point equivalence scoring system ranging from much inferior (-3) through equivalent (0) to much superior (+3) was used in each category. Radiation dose was recorded and compared with that in conventional measurements. Turnaround times of patients undergoing digital and conventional X-rays were evaluated. Clinical and radiographic issues were assessed by staff feedback. The digital images when compared with conventional film had an overall mean equivalence score of -0.429, with a standard deviation (SD) of 0.77. The best digital performance was in the mediastinum (mean 0.346, SD 0.49) and the weakest was for bony detail (mean -0.654, SD 0.81). Relative digital radiation dose compared to conventional varied from 72% (chest) to 2% (pelvis), with a simple average of 6%. Radiographic points included full-body imaging capability and differing positioning, penetration, workflow and practicality considerations. The digital images required overall patient times of 5-6 min, compared with 8-48 min for conventional X-rays. New installations are under way, and computed tomography and angiography applications are being explored. FDA approval is awaited. Projected cost is similar to that of flat-panel digital units. This digital unit was felt to be diagnostically substantially equivalent to conventional radiographs, with low-dose full-body imaging, improved workflow, digital technology and long-term cost benefits as potentially favourable contributions to trauma imaging.