We used immunological methods to determine cytosolic and nuclear steroid receptors to evaluate the advantages of nuclear receptor measurement in the selection of breast cancer patients for treatment. Around 75% of tumors showed coincidence between nuclear and cytosolic receptors (+/+ or –/–) for estrogen receptor (ER) and for progesterone receptor (PgR). Only cytosolic receptors were detected in around 20% of tumors. Distributed in the ER/PgR phenotypes according to the nuclear or cytosolic receptors, 64% of tumors remained in the same subgroup, whereas 16% of tumors were classified as hormone dependent according to cytosolic and independent according to nuclear receptors, which could be considered as ‘false-positive’ results. 6% of tumors would be classified as negative according to cytosolic receptors but positive according to nuclear receptors and would correspond to ‘false-negative’ results by conventional methods. Cytosolic receptor results may overrate the hormone dependence and cause some ‘misclassifications’ of patients. This could partially explain the lack of response to therapy in some cases.