Only the controlled trial method, clinical equivalent to the experimental method, with its successive phases and randomization, is able to confirm a real causal relationship and quantify the risk of error (alpha). However, the study must have sufficient power and randomization must not have resulted in an unbalanced distribution of various parameters likely to influence the result. Other methods, particularly surveys and case studies, only provide presumptions of causality. This review article, illustrated by three examples from the urological literature, is designed to demonstrate the difficulties of establishing a causal relationship when possible biases and confounding factors are taken into account.