Sentinel lymph node (SLN) biopsy, a new standard of care for staging invasive breast cancer, remains controversial for ductal carcinoma in situ (DCIS). Although DCIS has a natural history in which long-term distant disease-free survival (DDFS) is 98% to 99% and axillary node metastases are historically rare, three recent DCIS series have found SLN metastases in a surprising 6% to 13% of patients. The strongest argument for SLN biopsy in DCIS is the diagnostic uncertainty and inherent sampling error of conventional pathologic techniques. Definitive surgery (excision or mastectomy) reveals invasive cancer in 10% to 21% of patients with a preoperative core needle or surgical biopsy diagnosis of DCIS, all of whom become conventional candidates for SLN biopsy. In the absence of proven invasion, most positive SLN in DCIS are micrometastases detected by hematoxylin and eosin- and immunohistochemical-stained serial sections. An increasing body of evidence suggests that these are prognostically significant, not artifactual. We propose that DCIS patients with positive SLN have occult invasive cancers, and that the same may have been true for the 1% to 2% of DCIS patients who go on to develop distant metastasis, either after an invasive local recurrence or as a first event. We further suggest that the diagnosis of DCIS encompasses two patient populations: 1) a majority (perhaps 90%) with true in situ disease (or prognostically insignificant invasion), negative SLNs, and an expected DDFS of 100%; and 2) a minority (perhaps 10%) with occult invasion, positive SLNs, and an expected DDFS of perhaps 90%. Pending the development of predictive models for preoperative identification of this SLN-positive minority of DCIS patients, SLN biopsy is indicated in any DCIS patient who may have an underlying invasive cancer, especially those who require mastectomy. In DCIS, SLN biopsy may ultimately prove to be a more sensitive screening test for occult invasion than examination of the breast itself.