Bone marrow scintigrams (MS) were performed on 73 men with histologically proven prostatic carcinomas. Follow-up data were available in 24 cases. Thirty-six patients had skeletal metastases at the time of the first MS investigation. They were compared with conventional bone scintigrams (BS) as well as clinical, biological, radiological and other follow-up data obtained for the same patients. At the present stage of revision and of follow-up of our patients, MS appeared less sensitive than BS in diagnosing skeletal metastases (94% as against 100% if all abnormal MS and BS presentations are considered as diagnostic or 83% as against 86% if more restrictive analytic criteria were applied). On the other hand, MS showed slightly higher specificity (81% for MS as against 73% for BS) when all abnormal MS and BS presentations are considered as diagnostic. With more restrictive diagnostic criteria, BS remains more specific (100%) than MS (87%). Nevertheless, the likelihood of bone metastasis is low when a normal MS is found in case of dubious BS situations, but high when the MS is pathological. In 11 patients with skeletal metastasis for whom follow-up investigations were available, MS appeared better than (4 cases) or equivalent to (6 cases) BS for evaluating the responses and the ultimate evolution of the disease. Marrow scintigrams are thus a useful tool in cases of dubious BS and for evaluating the response of skeletal metastasis to treatment.