To determine the effects of 211At-labelled antibodies in solid tumour tissue, nude mice carrying OHS human osteosarcoma xenografts received intratumour injections at dosages of 1, 2 or 4 MBq (-1) tumour. The radioisotope was conjugated to either the osteosarcoma-specific monoclonal antibody TP-3 or the non-specific polyclonal antibody hlgGkappa. Tumour retention of injected radioimmunoconjugate (RIC), measured as the percentage of injected activity dosage per gram, was significantly higher for the [211At]TP-3 (203 +/- 93 at 24.1 h post injection) compared with the [211At]hlgGkappa (57 +/- 22 at 23.2 h post injection). The radioactive count rates in body (measured at neck and abdomen) were significantly lower with the TP-3 than with the hlgGkappa. Microautoradiography of the tumour radionuclide distribution was different for the two RICs, i.e. the [211At]TP-3 was to a larger extent concentrated near the injection site, whereas the [211At]hlgGkappa was more evenly distributed all over the tumour. The tumour growth was significantly delayed as a function of the injected activity dosage but without significant difference between the specific and the non-specific RIC. According to this study, it is possible to deliver highly selective radiation doses to solid tumours using intratumour injection of alpha-particle-emitting RICs. Improved tumour retention caused by antigen binding indicates that reduced normal tissue exposure can be obtained with antigen-specific antibodies. The heterogeneous tumour dose distribution observed is, however, a major impediment to the use of alpha-particle emitters against solid tumours.