In assisted reproduction, there is strong evidence for some things done, but no or only very weak evidence for others. There are several reasons for this. Most assisted reproduction procedures have small signal-to-noise ratios. This means that their treatment effect is sometimes only little better than the spontaneous conception rate, or the conception rate with traditional treatment. Hence, large trials are required. These demand complex multicentre logistics. The latter require substantial funding and funding for reproductive medicine in most countries is notoriously difficult to obtain (as opposed, for example, to oncology research or cardiovascular research). Apart from these funding issues, the creation of embryos specifically for research is only allowed in a limited number of European countries, thus tempting clinicians to skip preclinical studies altogether and go directly for clinical application in their patients, raising an ethical issue. Introducing new treatments into the clinic without proper evidence, however, is perhaps even more of an ethical issue. Subfertile couples are very vulnerable and should not be exploited.