The justification for performing surgery on patients most often rests on observational studies and rarely on correctly conducted randomised trials, which is the highest level of evidence. For methodological and ethical reasons, surgical trials can be problematic. However, the ethical problems are greater if trials are not performed, since it will then not be possible to find out which of the available interventions have no effect or are harmful. The problems can be overcome in many cases, and by intra- and international co-operation it would be possible to carry out many more trials, which would increase the level of evidence for surgery in daily practice. When randomised trials have not been performed, one is obliged to look for the evidence in prospective cohort studies, e.g. clinical databases, which, however, are more suitable for evaluating complications to surgery under routine circumstances than for evaluating useful effects.