BACKGROUND: NY-ESO-1 protein formulated in ISCOMATRIX™ results in CD4+, CD8+ T cell and antibody-mediated immunity. We evaluated persistence of immunity, relapse-free survival and tumour antigen expression upon relapse in patients vaccinated in an earlier trial. METHODS: Immunity was measured in 28 patients with resected NY-ESO-1-expressing tumours (melanoma 25, breast 3) 252-1,155 days (median = 681) after vaccination. In the earlier vaccination, trial patients received NY-ESO-1 with ISCOMATRIX™ adjuvant at three protein doses 10 μg, 30 μg or 100 μg (n = 14); 100 μg NY-ESO-1 protein (n = 8) or placebo (n = 6), together with 1 μg of intradermal (ID) NY-ESO-1 protein twice for DTH skin testing. Immune responses assessed in the current study included antibody titres, circulating NY-ESO-1-specific T cells and DTH reactivity 2 days after DTH skin testing with NY-ESO-1 protein (1 μg) or peptides (10 μg). Relapse-free survival was determined for 42 melanoma patients. On relapse NY-ESO-1 and HLA, class I was assessed by immunohistochemistry in 17. RESULTS: Persisting anti-NY-ESO-1 immunity was detected in 10/14 recipients who had previously received vaccine with ISCOMATRIX™ adjuvant. In contrast, immunity only persisted in 3/14 who received 100 μg un-adjuvanted NY-ESO-1 protein (3/8) or 2 μg DTH protein (0/6) P = 0.02. Hence, persisting NY-ESO-1 immunity was associated with prior adjuvant. Tumour NY-ESO-1 or HLA class I was downregulated in participants who relapsed suggesting immunoediting had occurred. CONCLUSION: Immunoediting suggests that a signal of anti-tumour activity was observed in high-risk resected melanoma patients vaccinated with NY-ESO-1/ISCOMATRIX™. This was associated with measurable persisting immunity in the majority of vaccinated subjects tested. A prospective randomised trial has been undertaken to confirm these results.