The purpose of this prospective study of sentinel lymph node (SLN) biopsy in a large series of melanoma patients with clinically negative regional lymph nodes from one cancer centre was to analyse the reliability of the procedure, the pattern of failures during follow-up and the factors affecting the clinical outcome of patients. Between April 1995 and November 2001, 726 consecutive patients with primary cutaneous malignant melanoma underwent SLN biopsy with preoperative lymphoscintigraphy. The vital blue dye technique was used in 170 patients, and the blue dye technique combined with intraoperative lymphoscintigraphy in 556 patients. The primary melanoma sites were head and neck in nine patients, the extremities in 419 patients, and the trunk in 298 patients. The median Breslow thickness was 3.0 mm. All patients were followed closely, the median follow-up time being 34 months. The sentinel node(s) were successfully identified in 96% of patients. Intraoperative lymphoscintigraphy combined with the blue dye technique improved the SLN identification rate (technical success in 97.3% of cases) compared with the blue dye technique alone (technical success in 91.6%). The rate of failed SLN procedures was significantly (P = 0.007) lower in inguinal basins (3.1%) compared with axillary basins (7.9%). SLN metastases were detected in 147 patients (20.2%). The presence of SLN metastases correlated significantly with primary tumour thickness and ulceration (P < 0.001). The false-negative SLN biopsy rate was 4.66% (27 out of 579 SLN-negative patients). All but two node-positive patients underwent complete lymphadenectomy. Lymph nodes other than SLNs were found to contain metastases in 26.9% of patients (39 out of 145). The 5 year overall survival (OS) rate was 84% for SLN-negative patients and 40% for SLN-positive patients. Five variables showed a strong, statistically significant negative independent prognostic association with OS: positive SLN status (P = 0.000001), primary melanoma thickness > 4 mm (P = 0.0009), male gender (P = 0.001), more than one lymph node involvement (P = 0.02) and lymph node extracapsular extension (P = 0.03). SLN biopsy is currently a valuable and effective diagnostic procedure for the precise staging of patients with clinically N0 cutaneous melanoma. So far SLN biopsy seems to be the only accessible method for consciously oriented detection of nodal micrometastases in melanoma that would otherwise go undetected. SLN status is the most important factor proven to distinguish high and low risk melanoma patients.