To investigate whether ultrasonographic joint assessment can predict the clinical response to intra-articular injection therapy of the knee. Patients with persistent gonarthritis intra-articularly received in a randomized double-blinded crossover fashion radiation synovectomy or a glucocorticoid injection, both followed by clinical bed rest. Prior to treatment and 3 months afterwards, grey-scale ultrasonography (US) of the knee was performed, measuring synovial thickness and extent of effusion. The final clinical effect of these two treatments was assessed at 3 months and finally at 6 months using a composite index. Ninety-seven patients, mainly suffering from undifferentiated arthritis (40%) or rheumatoid arthritis (31%), received 165 injections (including crossovers). Clinical effect at 6 months was not related to the baseline ultrasonographic extent of effusion or synovial thickness, nor with ultrasonographic decrease of effusion after the first 3 months. Nevertheless, it was associated with ultrasonographic decrease of synovial thickness within the first 3 months. Simple baseline US measurements fail to predict the final clinical effect of intra-articular treatment of the knee at 6 months, in contrast to early US changes of synovial thickness 3 months after therapy.