The current trend is to reserve the diagnosis of premature ejaculation (PE) for cases where penetration lasts for about one minute or less. The rationale is that the aetiology is primarily bio-constitutional, and that long-term pharmacological treatment is the only viable option. However, the literature contains little scientific evidence to support this argument. In fact, a good number of individuals who suffer from overly rapid ejaculation present with penetration duration exceeding one minute, and even severe forms of PE have responded favourably to psycho-sexological treatment. Moreover, although certain biological variables are known to influence ejaculation latency time, nothing indicates that they play an exclusive role of psychosocial etiological factors in severe PE. Therefore, it would be ‘premature’ to base a PE diagnosis on a maximum penetration duration of one minute, which should instead be considered a severity gradient. Given that desired criteria for penetration duration often exceed biological norms, it would be inappropriate to propose that only the most severe forms of PE have constitutional origins. In any case, the constitution is relatively flexible, and can respond to adaptive learning. An adaptive learning approach would undoubtedly be more difficult to apply in severe cases, but not impossible. The issue of whether to use pharmacological versus psycho-sexological treatment could be sidestepped by moving beyond the single criterion of ejaculation latency.