Randomized controlled clinical trials provide the best method to distinguish a drug from placebo without the inevitable selection biases that are seen in standard clinical care. This article reviews designs and limitations of clinical trials that are used in rheumatic diseases. The primary design in clinical trials is a parallel, in which patients are randomized in parallel to different therapies at different dosages or placebo. In recent years, other designs have been used increasingly, including "step-up," "step-down," and "cross-over" designs. Limitations of clinical trials in chronic diseases include a short time frame versus the long duration of disease, inclusion and exclusion criteria, use of surrogate markers that may not represent clinically relevant markers, statistical significance does not necessarily indicate clinical significance necessarily, and the fact that a control group does not assure the absence of bias. Therefore, long-term databases are needed to supplement clinical trials in analyzing results of therapy for rheumatoid arthritis.