OBJECTIVE Approximately 35% of patients with Graves' ophthalmopathy do not respond to immunosuppressive treatment. A possible explanation for this finding is that only patients with active ophthalmopathy respond to immunosuppressive treatment, whereas patients with fibrotic end stage disease do not. To distinguish between these two groups and to predict the outcome of immunosuppressive treatment, we developed a clinical activity score (CAS) based on four of the five classical signs of inflammation and tested its efficacy in a double‐blind, prospective study. DESIGN, PATIENTS AND MEASUREMENTS The CAS was determined by an ophthalmologist before, on the day of, and after the start of either oral prednisone or retrobulbar irradiation in 43 patients with moderate to severe Graves' ophthalmopathy. The therapeutic outcome was determined by a second ophthalmologist unaware of the CAS stores given. Success of treatment was defined as an improvement in NOSPECS class or grade. RESULTS Responders (22) and non‐responders (21) did not differ in age, sex, duration or severity of their Graves' ophthalmopathy. The pretreatment CAS, however, was significantly higher in responders than in non‐responders. Twelve of 22 responders and three of 21 non‐responders had a CAS ≥ 4 (80% vs 36%; P < 0.01). Using this CAS cut‐off point, the accuracy of CAS in predicting the therapeutic outcome was: specificity 86%, sensitivity 55%, positive predictive value 80%, negative predictive value 64%. Patients with a CAS ≥ 4 had a similar duration of Graves' ophthalmopathy as patients with a CAS < 4. CONCLUSIONS The clinical activity score has a high predictive value for the outcome of immunosuppressive treatment in Graves' ophthalmopathy. Disease activity, and not disease duration, is the prime determinant of therapeutic outcome.