Disclosed are methods and compositions for identifying malignant tumors that overexpress the c-erbB-2 oncogene. Assays useful for diagnosis and prognosis of neoplastic disease are provided which detect the external domain of c-erbB-2, the glycoprotein gp75 and quantitate the level of gp75 in the biological fluids of mammals carrying a tumor burden.Further disclosed are recombinant, synthetically and otherwise biologically produced novel proteins and polypeptides which are encoded by the external domain DNA sequence of the c-erbB-2 oncogene (the gp75 gene) or fragments thereof. Such gp75 proteins and polypeptides are useful as vaccines, therapeutically in the treatment of cancer either alone or in combination with chemotherapeutic agents.Also disclosed are antibodies to such gp75 proteins and polypeptides which are useful diagnostically and therapeutically. Still further disclosed are test kits embodying the assays of this invention.