Objectives-To describe (1) the characteristics of attenders to an urban accident and emergency (A&E) department over a one year period according to the frequency of their attendance and (2) the features of their attendances according to the frequency which the patient attended the department during the study period.Methods-A dataset containing information on all new attendances to an urban A&E department in 1995 was formed. For each attendance the following information was recorded: day of the week, hour of attendance, referral source, triage category, and disposal. A second dataset, consisting of the individuals ("attenders") who made the A&E visits ("attendances") during 1995 was also produced. For each attender the following information was recorded: age, sex, postal: code of residence, socioeconomic status, marital status, and number of attendances during 1995. A methodology reviewing time trends of frequency of attendance was utilised, as opposed to the use of an arbitrary cut off point.Results-34908 patients made 46735 visits in 1995. Increasing frequency of attendance was significantly associated with increasing age (Kruskal-Wallis <0.001), being male (chi(2) for linear trend 14.06, p<0.001), having a local postal address (chi(2) 279.79, p<0.0001), general medical services eligibility (chi(2) 781.67, p<0.0001), and inversely associated with being married (chi(2) 33.91, p<0.0001). Increasing frequency of attendance was significantly associated with attendance between the hours of 1700 and 0900 (chi(2) 295.62, p<0.001), being triaged as a nonemergency (chi(2) 1254.33, p<0.0001), and self referral (chi(2) 141.4, p<0.01001).Conclusions-A small group of A&E attenders accounts for a disproportionately large percentage of the total number of departmental attendances. The characteristics of frequent A&E attenders suggest that they may represent a vulnerable group of patients. A follow up study of the utilisation of all primary care services by such patients is suggested.