Lymph node status is still the single most important prognostic factor in breast cancer. Axillary surgery remains the only reliable means of providing this information. This pilot study evaluates using a highly specific radiolabelled monoclonal antibody to provide equivalent information by a non-invasive technique. After optimisation of labelling conditions, our first antibody, ICR12 (against the gene product of c-erbB-2) was evaluated in a mouse model system. Twenty-four hours post i.v. injection the mice were killed and their organs, blood and tumours harvested for counting. Tumour localisation was four times greater than that into normal tissues, reaching 20% injected dose per gram of tumour. Eight patients have had this Tc99m-ICR12. Patient selection was by immunocytochemical staining of fine needle aspirates from the patient's own breast cancer. After intravenous administration of the immunoconjugate, tomographic images were obtained at 24 h. These results were compared to the subsequent histopathological examinations. Three patients acted as normal controls, one patient was negative due to inappropriate sampling, and two patients had strong membrane staining and provided excellent tumour localisation to both breast primary and regional node metastases. A further two patients only had moderate antigen expression on staining and did not localise well. The good performance of this radiolabelled antibody with patients that strongly stain for the antigen encourages the development of this system as both a method of staging breast cancer and a potential means of immunotherapy in this subgroup of patients.