While diagnostic criteria for premenstrual syndromes (PMS) exist, studies rarely state how these criteria are operationally applied. We examined the consequences of application of different operational methods for DSM-IV criteria for premenstrual dysphoric disorder (PMDD) to individual cycles in women with PMS and controls. PMDD criteria require the presence of both certain types or numbers of symptoms (5/11 symptoms present premenstrually, at least one being one of four mood symptoms) and certain phenomenal characteristics (present premenstrually, absent postmenstrually, causing interference premenstrually). We identified individual cycles as symptomatic or asymptomatic by applying criteria that operationalized the required phenomenal elements of PMDD according to four severity thresholds: literal (i.e. present or absent), 30%, 50%, 70%. Data examined were Daily Rating Form symptom scores from two symptomatic menstrual cycles both in 25 women with PMS and 25 controls. Literal thresholds correctly identified 28% of symptomatic and 4% of asymptomatic cycles, compared with 86 and 70% identification with a 30% threshold, 60 and 86% with a 50% threshold, and 0 and 100% with a 70% threshold. An "optimal" combination of 30% thresholds for premenstrual symptomatology and premenstrual interference and a 50% threshold for postmenstrual symptomatology correctly identified 92% of symptomatic cycles in women with PMS and 72% of asymptomatic cycles in controls. Different criteria for cycle inclusion yield markedly different sample compositions. No single operational threshold of the phenomenal features maximizes selection of both symptomatic and asymptomatic cycles, largely consequent to the ubiquity of postmenstrual symptoms and premenstrual "interference" even in women without PMS. At the very least, the method for operationalizing DSM-IV criteria should be described in studies of PMDD.