This article aims to clarify why, and by which mechanisms, exercise may influence the normal menstrual cycle. Therefore, the vast amount of literature on this subject is reviewed and a critical appraisal of the most widespread hypotheses if offered. The strikingly low body mass which frequently accompanies exercise-related menstrual irregularities (ERMI) has led some authors to develop a hypothesis which postulates that a critical percentage of body fat is essential to trigger normal menstruation. The relevance of any reference to anorexia nervosa to support this view lacks consistency: female athletes differ in many ways from patients with anorexia nervosa, not least in their excellent physical status which is essential to deliver first-class performances. ERMI is not identical to the so-called female athlete triad, a complicated pathology that involves ERMI, premature osteoporosis and disordered eating. ERMI itself does not seem to have any substantial pathological effects as long as attention is paid to preventing osteoporosis or stress fractures which may result from prolonged hypo-estrogenaemia. In the female athlete with ERMI who wishes to conceive, the accompanying subfertility may necessitate a response other than a prompt reduction in training intensity, as this is hardly a first choice for any top athlete. During recent years, a number of prospective studies have greatly contributed to our understanding of the complexity of the mechanisms involved in ERMI. Older hypotheses, such as those considering hyperprolactinaemia as the cornerstone of ERMI, have now been firmly rejected. The present hypotheses emphasise the importance of caloric deficiency and limited energy availability, although they still fail to identify the actual mechanism that causes ERMI. There is, however, evidence that ERMI is produced by a disturbance of the hypothalamic gonadotrophin-releasing hormone oscillator. This disturbance is caused by either an insufficient estrogen or progesterone feedback or by an imbalance of local opioid peptide and catecholamine activities mediated by gamma-aminobutyric acid (GABA), corticotrophin-releasing hormone and insulin-like growth factor-1. More recent experiments have also linked ERMI with changes in steroid metabolism, in particular, an increasing activity of catecholestrogens possibly leading to enhanced intracerebral noradrenaline (norepinephrine) levels that may interfere with normal gonadotrophin release. This article demonstrates that the outcome of the many studies of ERMI is characterised by much controversy and numerous methodological flaws. The importance and complexity of some recent findings necessitate a comprehensive study which links older and newer findings within a critical perspective.