Researchers investigating premature ejaculation (PE) have employed widely diverse definitions of it, thereby limiting progress in the field. This study summarizes available research on PE, notes patterns that emerge from these studies, compares patterns across several types of studies, and suggests a common model for defining PE groups to guide future research. We surveyed two bibliographic databases, identifying 45 studies employing a definition or description of a PE group. From these, we extracted a range of information, including descriptions of the participants, recruitment procedures, if PE subtypes were identified, operational criteria for PE classification, relationship and partner information, and additional inclusion/exclusion criteria. Over 50% of studies reported no criteria, or relied on simple self-identification by participants to establish the PE group. Quantifiable behavioral criteria were used in 49% of the studies, with ejaculatory latency reported most frequently. This measure was also used as a criterion more frequently in studies focusing on assessment of sexual response, whereas the number of penile thrusts was used more frequently in studies prior to 1989. Partner information was often included but seldom used as part of the assessment procedure. Progress on research and treatment of PE will continue to be limited by the absence of commonly accepted criteria for PE group membership and by a lack of identification of relevant PE subtypes and etiologies. This paper suggests a flowchart, based on data and a rational analysis of 40 years of research, for characterizing PE in ways that could assist the development of the field.