Background The premature ejaculation diagnostic tool (PEDT) is a brief, multidimensional validated instrument devised for diagnosing premature ejaculation (PE). However, there is insufficient evidence regarding its ability to differentiate subgroups of PE. We assessed the ability of the PEDT to differentiate four subgroups of PE (lifelong, acquired, variable and subjective PE). Methods: An internet-based survey was conducted with a population-based sample of males aged 20–59 years. Participants were asked to complete a questionnaire requesting detailed medical and sexual histories. The questionnaire including questions from the PEDT and from the Medical Outcome Study Short-form 36-Item Health Survey (SF-36). Using the PEDT, PE was defined as a cutoff score of ≥11. Results: In this study, 443 subjects (mean age 39.3 ± 10.1 years) were included. PEDT-PE prevalence was 14.6%. The proportions of PE subgroups and their mean PEDT scores were: lifelong PE, 2.9% and 15.5; acquired PE, 7.0% and 11.2; variable PE, 7.4% and 10.4; subjective PE, 3.2% and 9.0. PEDT scores were significantly higher in the lifelong PE group than in other the subgroups of PE (P < 0.001). The subjective PE group had the lowest PEDT score, and their physical and mental component scores for the SF-36 were similar to those of non-PE subjects. Conclusions: This population-based cross-sectional survey has demonstrated that the PEDT is not appropriate for research into four subgroups of PE when used in an general male population study, unless the PEDT is combined with an additional questionnaire with specific questions on the four PE subgroups.