In assessing the benefits of health services it is presumed that the relationship of cause and effect can be determined by scientific methods. In general, the randomized controlled trial (RCT) is considered the gold standard for the generation of scientific evidence. Yet there is an increasing amount of data indicating that not in all situations RCTs seem to be the adequate tool for generating evidence. In many instances, RCTs are, in fact, barely useful to reflect the real-world situation of health-care services. Some of the well-known yet mostly ignored limitations of RCTs are summarized.The absence of definite a priori preferences for a specific kind of intervention (equipoise) is an essential prerequisite for both physicians' and patients' consent to participate in an RCT. Numerous examples of quantitative studies confirm that the willingness to accept randomization is limited in operative disciplines. If it is true that the invasiveness of a diagnostic or therapeutic intervention correlates with the preferences of doctors and patients, the small number of RCTs in operative fields should be expected. Further development of the important concept of RCTs should specify the conditions under which RCTs can generate significant results. Paying attention to this will open up new perspectives for the assessment of health-care services.