Sentinel node biopsy allows an accurate selection of melanoma patients to be submitted to therapeutic dissection. From February 1994 to August 1998, at the National Cancer Institute, S. Pio X Hospital in Milan and Bufalini Hospital in Cesena, 580 sentinel node biopsies were performed in 540 stage I melanoma patients (242 males; 298 females; median age 47). Primary melanoma was located in the trunk in 201 patients, in lower limbs in 242 cases, in upper limbs in 80 cases and in head and neck in 17 patients. Injection of blue dye for sentinel node identification was performed in all cases; 372 patients were submitted to preoperative lymphoscintigraphy and in 272 cases an intraoperatory probe for a radioguided biopsy was utilized. Sentinel node identification rate was 91%. Sentinel node positivity rate was 15%. Frozen sections were examined in 199 cases. Distribution of positive cases according to primary thickness is the following: <1 mm: 1%; 1-1.99 mm: 5%; 2-2.99 mm: 18% and > or =3 mm: 27%. Sentinel node appeared to be the only metastatic node in 77% of patients submitted to dissection. The adoption of preoperative lymphoscintigraphy and the intraoperative use of the gamma probe contributed substantially in S.N. identification. No complications caused by the procedure were reported. Eight patients had a regional node relapse after a negative sentinel node biopsy and were submitted to therapeutic distant dissection. Currently 513 patients are alive with no evidence of disease. Present data confirm the feasibility and safety of sentinel node technique for selection of patients to be submitted to radical node dissection and to eventual adjuvant treatments.