With respect to recurrent pregnancy loss (RPL), unfortunately there is very little consensus about which investigations are useful for identifying causes and evaluating the prognosis, and also about which treatments are effective. In this review, arguments are given for the claim that this lack of consensus may mainly be because studies in the field of RPL have yielded very heterogeneous results. This heterogeneity, in the authors' belief, is caused by the scientists' lack of appreciation, when they do research, of important epidemiological knowledge about RPL (e.g., about the multifactorial background for most of the RPL cases and the importance of matching/adjusting for a series of prognostic variables when groups are mutually compared). Furthermore, many studies in RPL contain methodological flaws that are sometimes severe. A series of important epidemiological features of RPL is highlighted in the review and the most important methodological pitfalls, many of them specific for RPL research, are discussed. Advice is given about to how to avoid the pitfalls in order that the validity of the studies can improve for the benefit of the patients.