OBJECTIVE: Hand osteoarthritis is a prevalent disease with limited treatment options. Since joint inflammation is often present, we investigated tumour necrosis factor (TNF) as treatment target in patients with proven joint inflammation in a proof-of-concept study. METHODS: This 1-year, double-blind, randomised, multicentre trial (NTR1192) enrolled patients with symptomatic erosive inflammatory hand osteoarthritis. Patients flaring after non-steroidal anti-inflammatory drug washout were randomised to etanercept (24 weeks 50 mg/week, thereafter 25  mg/week) or placebo. The primary outcome was Visual Analogue Scale (VAS) pain at 24 weeks. Secondary outcomes included clinical and imaging outcomes (radiographs scored using Ghent University Scoring System (GUSS, n=54) and MRIs (n=20)). RESULTS: Of 90 patients randomised to etanercept (n=45) or placebo (n=45), respectively, 12 and 10 discontinued prematurely. More patients on placebo discontinued due to inefficacy (6 vs 3), but fewer due to adverse effects (1 vs 6). The mean between-group difference (MD) in VAS pain was not statistically significantly different (-5.7 (95% CI -15.9 to 4.5), p=0.27 at 24 weeks; - 8.5 (95% CI -18.6 to 1.6), p=0.10 at 1  year; favouring etanercept). In prespecified per-protocol analyses of completers with pain and inflammation at baseline (n=61), MD was -11.8 (95% CI -23.0 to -0.5) (p=0.04) at 1  year. Etanercept-treated joints showed more radiographic remodelling (delta GUSS: MD 2.9 (95% CI 0.5 to 5.4), p=0.02) and less MRI bone marrow lesions (MD -0.22 (95% CI -0.35 to -0.09), p = 0.001); this was more pronounced in joints with baseline inflammation. CONCLUSION: Anti-TNF did not relieve pain effectively after 24 weeks in erosive osteoarthritis. Small subgroup analyses showed a signal for effects on subchondral bone in actively inflamed joints, but future studies to confirm this are warranted.