Based on a 2-year controlled double-blind trial of levamisole, penicillamine, and azathioprine (L, P, and A), a computer aided search for predictive factors of outcome was instituted. Already at month 4 several indicators of synovitis activity were able to discriminate between patients staying in the trial for 24 months and patients whose treatment was discontinued before that time. Patients who had previously received gold therapy responded less favourably to L, P, and A than those who had not received gold. This reduction of response was more pronounced in gold resistant patients than in patients whose gold treatment had been discontinued for other reasons. The only phase protein (of several) with a predictive value was haptoglobin. If, after 4 months of treatment, haptoglobin did not normalize, this finding indicated a lack of response to treatment or a deterioration of synovitis activity during the following 4 months. The response to treatment was not influenced by HLA-types, sex, age, or clinical synovitis, disease duration, functional or anatomical aberrations at the start of treatment. The shape of the response curve as reflected by means of monthly measurements of serum-albumin and ESR was not related to disease duration, HLA-types, or previous gold treatment.