The use of elective regional node dissection in patients with cutaneous melanoma without any clinical evidence of metastatic spread is still debated. Our aim was to evaluate the efficacy of immediate node dissection in patients with melanoma of the trunk and without clinical evidence of regional node and distant metastases. An international multicentre randomised trial was carried out by the WHO Melanoma Programme from 1982 to 1989. The trial included only patients with a trunk melanoma 1.5 mm or more in thickness. After wide excision of primary melanoma, patients were randomised to either immediate regional node dissection or a regional node dissection delayed until appearance of regional-node metastases. Of the 252 patients entered, 240 (95%) were eligible and evaluable for analysis. 122 of these were randomised to immediate node dissection. 5-year survival observed in patients who had delayed node dissection was 51.3% (95% CI 41.7-60.1) compared with 61.7% (52.0-70.1) of patients who had immediate node dissection (p=0.09). 5-year survival rate in patients with occult regional node metastases was 48.2% (28.0-65.8) and 26.6% (13.4-41.8, p=0.04) in patients in whom the regional node dissection was delayed until the time of appearance of regional node metastases. Multivariate analysis showed that routine use of immediate node dissection had no impact on survival (hazard ratio 0.72, 95% CI 0.5-1.02), whilst the status of regional nodes affected survival significantly (p=0.007). The patients with regional nodes that became clinically and histologically positive during follow-up had the poorest prognosis. Node dissection offers increased survival in patients with node metastases only. Sentinel node biopsy may become a tool to identify patients with occult node metastases, who could then undergo node dissection.