Background. Glutathione (GSH) is important in in vitro models of radiation and drug resistance, but its clinical relevance is uncertain. GSH levels were measured prospectively in 35 patients with breast cancer to establish normal ranges and to determine whether differences exist between tumor, lymph node metastases (when present), and normal breast tissue. Methods. Fresh tissue was obtained immediately from discarded surgical specimens, and total GSH levels were measured with the Tietze cyclic reduction assay. When possible, multiple sites were assayed in each tumor to assess intratumor variability. Results. GSH levels in primary breast tumors were more than twice the levels found in normal breast tissue, and levels in lymph node metastases were more than four times the levels in normal breast tissue. The levels of GSH in normal lymph nodes were similar to levels in normal breast tissue. GSH levels in different areas of the same breast tumor ranged from below those of normal breast tissue up to 11 times normal tissue levels. No correlation was found between tumor GSH levels and common clinical parameters such as tumor size, nodal status, stage, estrogen receptor levels, or progesterone receptor levels. Conclusions. GSH appears to be a marker of breast malignancy that is independent of hormonal receptor status and stage and may be a marker of cells with increased potential for dissemination. Because of tumor heterogeneity, multiple sites should be sampled whenever possible. Long‐term clinical follow‐up will be necessary to ascertain the usefulness of tumor and lymph node GSH levels in predicting prognosis.