ABSTRACT Background In former days, information obtained from randomized well-controlled clinical trials and epidemiological studies on premature ejaculation (PE) was not available, thereby hampering the efforts of the consecutive DSM Work Groups on Sexual Disorders to formulate an evidence-based definition of PE. The current DSM-IV-TR definition of PE is still nonevidence based. In addition, the requirement that persistent self-perceived PE, distress, and interpersonal difficulties, in absence of a quantified ejaculation time, are necessary to establish the diagnosis remains disputable. Aim To investigate the validity and reliability of DSM and ICD diagnosis of premature ejaculation. Methods The historical development of DSM and ICD classification of mental disorders is critically reviewed, and two studies using the DSM-IV-TR definition of PE is critically reanalyzed. Results Reanalysis of two studies using the DSM-IV-TR definition of PE has shown that DSM-diagnosed PE can be accompanied by long intravaginal ejaculation latency time (IELT) values. The reanalysis revealed a low positive predictive value for the DSM-IV-TR definition when used as a diagnostic test. A similar situation pertains to the American Urological Association (AUA) definition of PE, which is practically a copy of the DSM-IV-TR definition. Conclusion It should be emphasized that any evidence-based definition of PE needs objectively collected patient-reported outcome (PRO) data from epidemiological studies, as well as reproducible quantifications of the IELT.