Abstract Objectives The value of US-defined tenosynovitis in predicting the persistence of inflammatory arthritis is not well described. In particular, the predictive utility of US-defined tenosynovitis of larger tendons is yet to be reported. We assessed the value of US-defined tenosynovitis alongside US-defined synovitis and clinical and serological variables in predicting persistent arthritis in an inception cohort of DMARD-naïve patients with early arthritis. Methods One hundred and fifty DMARD-naïve patients with clinically apparent synovitis of one or more joints and a symptom duration of ≤3 months underwent baseline clinical, laboratory and US (of 19 bilateral joints and 16 bilateral tendon compartments) assessments. Outcomes were classified as persistent or resolving arthritis after 18 months’ follow-up. The predictive value of US-defined tenosynovitis for persistent arthritis was compared with those of US-defined synovitis, and clinical and serological variables. Results At 18 months, 99 patients (66%) had developed persistent arthritis and 51 patients (34%) had resolving disease. Multivariate logistic regression analysis showed that US-detected digit flexor tenosynovitis odds ratio (OR): 6.6, 95% CI: 2.0 , 22.1, P = 0.002 provided independent predictive data for persistence over and above the presence of US-detected joint synovitis and RF antibodies. In the RF/ACPA-negative subcohort, US-defined digit flexor tenosynovitis remained a significant predictive variable (OR: 4.7, 95% CI: 1.4, 15.8, P = 0.012), even after adjusting for US-defined joint synovitis. Conclusion US-defined tenosynovitis provided independent predictive data for the development of persistent arthritis. The predictive role of US-defined digit flexor tenosynovitis should be further assessed; investigators should consider including this tendon site as a candidate variable when designing imaging-based predictive algorithms for persistent inflammatory arthritis development.