The aim of the study was to compare, in a randomized prospective study, the efficacy and safety of intraorbital administration of low doses of RTX versus intravenous glucocorticoids (GCs) to treat patients affected by moderately severe thyroid-associated active orbitopathy. Twenty patients with active, moderately severe TAO, whose mean age was 56.7 years±10.2 SD participated in the study. Patients were randomly selected and treated with intraorbital injections of RTX or with i.v. GCs. Disease activity and severity were assessed by the Clinical Activity Score (CAS) and the NOSPECS. Computed tomography or magnetic resonance scans were performed in all patients. In the RTX group, full blood cell count and flow cytometric analysis on peripheral blood lymphocytes were done. The patients were followed for 20 months. In both groups, CAS and NOSPECS indexes were significantly reduced (p<0.005). In particular, CAS reduction was evident since the first follow-up with both treatments. Proptosis decreased significantly only in group B and diplopia showed no significant changes during follow-up times in both groups. Neither of the treatments affected the peripheral TRab. In group A, 5 weeks after the first injection, the CD20+ peripheral lymphocytes value was nearly zero. One patient treated with rituximab progressed to severe TAO (optic neuropathy) following the second injection so the treatment was discontinued. The data confirm the therapeutic efficacy of RTX in active TAO, even in low doses and locally administered. The efficacy on the inflammatory component of the disease is comparable to that of steroids and seems to be related with the reduction of peripheral CD20+ lymphocytes. Caution should be given to an accurate patient selection.