Selective sentinel lymphadenectomy dissection has been demonstrated to have high predictive value for axillary staging in breast cancer patients. Preoperative lymphoscintigraphy can localize and facilitate the harvesting of sentinel lymph nodes (SNLs) with a high success rate. The failure rate of selective sentinel lymphadenectomy ranges between 2% and 8%. Details of the failures were seldom addressed. This study analyzes the causes of failure to harvest SLNs in spite of positive preoperative lymphoscintigraphy. From November 1997 through November 2000, 201 female patients with histologically confirmed and operable breast carcinoma underwent selective sentinel lymphadenectomy at the University of California, San Francisco (UCSF) Carol Franc Buck Breast Care Center. Among these patients, 183 (91%) received preoperative lymphoscintigraphy to identify axillary lymph nodes. The causes of failure to harvest the SLNs in this group of patients despite successful preoperative lymphoscintigraphy were analyzed. In our series, the failure rate of SLN identification was 7.0% (14/201). The failure rate for our first year was 11.1% (6/54), second year 9.1% (7/77), and third year 1.4% (1/70). The incidence of failure in spite of positive preoperative lymphoscintigraphy was 3.5% (6/170). The shine-through effect of the primary injection site and failure to visualize a blue lymph node were the main reasons for technical failure. Most of these cases occurred during our learning curve of the procedure. The possibility of failure to get the SLN should be explained to patients before surgery. Axillary lymph node dissection (ALND) should be done if selective SLN dissection is not successful.