In 63 patients with primary extramedullary malignant lymphoma or plasmacytoma, a study was performed in order to evaluate bone marrow involvement. All patients underwent a 99mTc microcolloid bone marrow whole body imaging (scintigraphy), using a gamma camera interfaced with a computer, followed by nuclear magnetic resonance bone marrow imaging (MRI), (1.5 Tesla). MR images were made of the lumbosacral region, the pelvic region, both femoral and other parts of the skeleton, according to focal lesions in the scintigraphy. A posterior iliac crest bone marrow biopsy was used as a standard reference. In the present study, both scintigraphy and MRI showed a dissiminated or focal involvement or a combination of both. In 53 of the 63 patients (84%) the results were in accordance. Pathological MR signals or pathological findings in scintigraphy did not always correspond to tumorous bone marrow involvement, and were shown to reflect reactive changes in the central part of the skeleton in combination with a periphery radionuclide extention interpreted as a periphery compensatory hematopoietic proliferation. The negative predictive value of scintigraphy and MRI was 92% and 100%, respectively. When combining the results of both examinations, the positive predictive value increased from 49% to 58%, if the bone marrow biopsy is accepted as gold standard. The results indicate that bone marrow investigation performed simultaneously using scintigraphy and MRI is superior both to the use of either of the methods alone and to the traditional iliac crest bone marrow biopsy.