In an attempt to compare the sensitivity of bone radiographs and bone marrow magnetic resonance (MR) imaging for bone lesion detection in patients with stage III multiple myeloma (MM) and to evaluate the possible consequences of the replacement of the conventional radiographic skeletal survey (RSS) by an MR survey of the spinal and pelvic bone marrow in these patients, we obtained MR studies of the thoracic and lumbar spine, pelvis and proximal femurs in addition to the conventional RSS (including radiographs of the skull, entire spine, pelvis, ribs, humerus and femurs) in 80 consecutive patients with newly diagnosed stage III MM according to the Durie and Salmon staging system (based on blood tests and on the RSS). The performance of MR and radiographic studies to detect bone lesions in given anatomic areas and in given patients were compared. The consequences on MM staging following the substitution of the RSS by the MR survey were assessed. MR imaging was superior to radiographs for lesion detection in the spine (76% v 42% of patients) and pelvis (75% v 46% of patients). The RSS was superior to the limited MR imaging survey for the detection of bone involvement in the patient population (87.5% v 79% of patients). If the RSS had been replaced by the MR imaging survey for patient staging, 7/80 patients would have been categorized as stage I and one as stage II MM on the basis of normal MR findings and biological findings consistent with these stages. Substitution of the RSS by a limited spinal and pelvic marrow MR survey would lead to 'understaging' of 10% of patients with otherwise stage III MM on the basis of blood tests and the conventional RSS.