We analysed the localisations of sentinel nodes (SN) found with the SN procedure to compare these sites with those that would have been predicted by conventional clinical descriptions of cutaneous lymphatic drainage. We assessed the surplus value of performing the SN procedure in melanoma patients who underwent regional nodal surgery. The SN procedure was performed in 348 patients with melanomas who were referred to our institute between 1993 and 1999. The localisations of the melanomas with the corresponding SNs were meticulously recorded on drawings of the human body and grouped according to the conventional descriptions. Predictability of lymph drainage was defined as the percentage of melanomas whose draining pattern was to the ipsilateral nearest basin, without simultaneous drainage to other basins or to an interval node. In all patients the SN procedure visualised at least one SN. We found 410 lymphatic basins in 347 patients. These basins included basins that could not have been predicted by the conventional clinical descriptions, such as multiple basins and contralateral drainage sites. For the head/neck region, SNs could be found in any of the basins described in the literature. The trunk's drainage predictability depended strongly on the melanoma localisation, ranging from 0% in the midline to 92% in one of the upper quadrants. The lower extremities had a high predictability of almost 100%, and predictability of drainage for the upper extremities ranged from 77% to 100%. In total, 34% of the patients had a cutaneous lymphatic drainage that was unpredictable, either totally or partially. We therefore conclude that an SN procedure is indispensable if the drainage site(s) are to be accurately identified.