Objective: To study the adherence of rheumatologists to the Dutch guidelines for anti-tumour necrosis factor alpha (TNF-α) treatment. The secondary objective was to evaluate alternatives to the present guidelines with regard to the percentage of responders and costs. Methods: The response (>1.2 DAS28 decrease) in patients who started on anti-TNF-α treatment for the first time was evaluated at 3 and 6 months after initiation. How many patients continued or discontinued their initial anti-TNF-α treatment was evaluated. Possible alternative guidelines were evaluated by means of a decision tree, with regard to the expected percentage of successfully (responders) and unsuccessfully treated patients and expected costs. Results: At 3 months 56% (N = 306) and 44% (N = 233) of all 539 evaluable patients were classified as responders or non-responders, respectively. Despite the guidelines, most (81%) (N = 189) of the non-responders continued treatment. 37% of the non-responders who continued anti-TNF-α treatment were eventually classified as responders at 6 months. Decision analytical modelling showed that with equal expected costs all alternative strategies would result in more responders than according to theoretical full adherence with the guidelines. “Continuation in case of partial response” had the best trade-off between successfully treated patients (64%) and unsuccessfully treated patients (17%). Conclusion: There was suboptimal adherence to the Dutch guidelines for treatment with anti-TNF-α for rheumatoid arthritis patients. This seemed to be justified by the fact that a delayed response up to 6 months was shown. If treatment is continued despite a non-response at 3 months, this is only recommended in patients with at least a partial response (at least 0.6 DAS28 improvement).