OBJECTIVES: X-rays are frequently used as the initial imaging study for low back pain (LBP). X-rays are inexpensive, but neither sensitive nor specific for many causes of LBP. Recently developed rapid magnetic resonance (RMR) imaging protocols provide more accurate anatomical information. Furthermore, because of reduced imaging time, RMR cost may approach that of X-ray. We conducted a randomized controlled trial measuring the societal perspective cost-effectiveness of RMR versus X-ray for patients with LBP. METHODS: Patients with lumbar x-rays ordered in one of four clinical centers were eligible. Consenting subjects were randomized to X-ray or RMR. The primary outcome was functional status (Roland scale) at 12 months. Economic outcomes included time trade-off (TTO), visual analogue scale (VAS), and SF-36 derived utility. Costs of RMR and X-ray imaging were determined in time-motion studies. Other health care and non-health care costs were measured using medical record review and questionnaires and valued using standard unit costs. RESULTS: 380 patients were randomized. Functional outcomes and cost data were available for 328 (86%) and 362 (95%) patients respectively. At 12 months the mean function of both groups had improved, but there was no difference between groups (X-ray 4.0 versus RMR 3.9; p=0.939). Likewise, there was no effect on VAS or SF-36 derived utilities. RMR patients had greater improvement in TTO scores (10.9 versus 3.2; p=0.026). Health care costs following RMR were higher with borderline significance ($2154 versus $1638; p=0.075), principally due to more specialist consultations and surgery. X-ray dominated RMR in the primary analysis. CONCLUSIONS: Our results suggest that the extra cost of RMR does not result in improved functional status and currently it should not replace X-ray in clinical practice. Given the poor correlation between TTO and other outcomes, the difference in TTO outcomes is probably a spurious finding but warrants further study.