The most striking feature of the history of treatment of patients with cirrhosis and ascites is the recurring cycle of initial enthusiasm for a new modality based on uncontrolled observations, followed by reports of complications and failures and/or negative randomized control trials (RCTs). The RCTs tend to be performed rather late, after it is realized that there are problems with the new treatment. In 1975 Tom Chalmers made a plea for randomization of the first patient treated with a new modality. The appropriateness of performing RCTs very early in the evaluation of a new treatment cannot be overemphasized today. Carefully designed RCTs that focus on appropriate subsets of patients and evaluate clinically important endpoints (rather than easier-to-measure, but unimportant indirect endpoints) are the keys to "evidence-based medicine" that will lead to the best outcomes for our patients. If we do not remember that uncontrolled studies regularly lead us into years or even decades of "blind alleys" of investigation, we are destined to repeat the mistakes of the past.