OBJECTIVES: To study whether there is an association between body mass index (BMI) category and survival of various tumour necrosis factor inhibitors (TNFi) in rheumatoid arthritis (RA) patients in a real-life longitudinal international registry. METHODS: Data from 5230 patients with RA starting treatment with any TNFi were selected from the METEOR registry. Patients were divided into six BMI categories: 3.7% underweight, BMI < 18.5 kg/m(2); 46% normal weight, BMI 18.5-25 kg/m(2); 32% pre-obesity, BMI 25-30 kg/m(2); 13% obesity class I, BMI 30-35 kg/m(2); 3.4% obesity class II, BMI 35-40 kg/m(2); and 1.6% obesity class III, BMI > 40 kg/m(2). Time on treatment in the different BMI categories was compared for all TNFi combined and for the infliximab, adalimumab and etanercept separately, using Kaplan-Meier curves and Cox regression analyses. Cox regression analyses were adjusted for potential confounders, with follow-up censored at 5000 days. RESULTS: Patients in obesity class II (HR 1.28, 95% CI 1.06 to 1.54) and III (HR 1.67, 95% CI 1.29 to 2.18) and underweight patients (HR 1.30, 95% CI 1.07 to 1.58) showed statistically significantly shorter TNFi survival than normal weight patients. The effect in underweight patients was strongest for infliximab (HR 1.82, 95% CI 1.20 to 2.76), the effect in overweight patients was strongest for infliximab (category II (HR 1.49, 95% CI 0.98 to 2.26); category III (HR 1.46, 95% CI 0.79 to 2.71)) and etanercept (category II (HR 1.27 95% CI 0.98 to 1.65); category III (HR 1.79, 95% CI 1.25 to 2.55)). No significant effect modification from reported pain was found. CONCLUSION: Both underweight and overweight patients discontinued TNFi treatment earlier than normal weight patients, without evidence of reported pain as the main determinant. It remains uncertain what determines TNFi survival in individual patients.