The effects of sorafenib–an oral multikinase inhibitor targeting the tumour and tumour vasculature–were evaluated in patients with advanced melanoma enrolled in a large multidisease Phase II randomised discontinuation trial (RDT). Enrolled patients received a 12-week run-in of sorafenib 400 mg twice daily (bid). Patients with changes in bi-dimensional tumour measurements< 25% from baseline were then randomised to sorafenib or placebo for a further 12 weeks (ie to week 24). Patients with⩾ 25% tumour shrinkage after the run-in continued on open-label sorafenib, whereas those with⩾ 25% tumour growth discontinued treatment. This analysis focussed on secondary RDT end points: changes in bi-dimensional tumour measurements from baseline after 12 weeks and overall tumour responses (WHO criteria) at week 24, progression-free survival (PFS), safety and biomarkers (BRAF, KRAS and NRAS mutational status). Of 37 melanoma patients treated during the run-in phase, 34 were evaluable for response: one had⩾ 25% tumour shrinkage and remained on open-label sorafenib; six (16%) had< 25% tumour growth and were randomised (placebo, n= 3; sorafenib, n= 3); and 27 had⩾ 25% tumour growth and discontinued. All three randomised sorafenib patients progressed by week 24; one remained on sorafenib for symptomatic relief. All three placebo patients progressed by week-24 and were re-started on sorafenib; one experienced disease re-stabilisation. Overall, the confirmed best responses for each of the 37 melanoma patients who received sorafenib were 19% stable disease (SD)(ie n= 1 open-label; n= 6 randomised), 62%(n= 23) progressive disease (PD) and 19%(n= 7) unevaluable. The overall median PFS was 11 weeks. The six randomised patients with SD had overall PFS values ranging from 16 to 34 weeks. The most common drug-related adverse events were dermatological (eg rash/desquamation, 51%; hand-foot skin reaction, 35%). There was no relationship between V600E BRAF status and disease stability. DNA was extracted from the biopsies of 17/22 patients. Six had V600E-positive tumours (n= 4 had PD; n= 1 had SD; n= 1 unevaluable for response), and 11 had tumours containing wild-type BRAF (n= 9 PD; n= 1 SD; n= 1 unevaluable for response). In conclusion, sorafenib is well tolerated but has little or no antitumour activity in advanced melanoma patients as a single agent at the dose evaluated (400 mg bid). Ongoing trials in advanced melanoma are evaluating sorafenib combination therapies.