OBJECTIVE: To determine if screening for undiagnosed type 2 diabetes mellitus (T2DM) and pre-diabetes is feasible in an Australian ED; to estimate the prevalence of T2DM and pre-diabetes in the Australian ED population. METHODS: Prospective cross-sectional prevalence survey in the ED of St Vincent's Hospital, Melbourne, an adult, tertiary referral centre seeing approximately 40,000 patients annually. A convenience sample of adult patients was screened with finger-prick random blood glucose and glycosylated haemoglobin (HbA1c); those over 6.0 mmol/L and 6.0% were referred for oral glucose tolerance test (OGTT). Diagnoses of T2DM and pre-diabetes were made according to World Health Organization definitions. Those not attending for OGTT were contacted by phone, and interviewed about their reasons. RESULTS: Seven hundred and twenty-five patients were recruited; 135 (18.6%; 95% confidence intervals [CI] 15.9-21.6%) had known T2DM, leaving 590 screened, of whom 210 screened positive. Of the 192 referred for OGTT, 147 (77%) did not attend despite several telephone reminders. Of the 45 (23%) completing OGTT, pre-diabetes was present in eight (17.8%; 95% CI 9.0-31.6%) and T2DM in six (13.3%; 95% CI 5.9-26.6%). Many people interviewed (18/86, 21%) did not attend for OGTT on the advice of their doctors. CONCLUSIONS: This inner city tertiary ED has a high prevalence of T2DM, diagnosed and undiagnosed, with over a quarter of our population probably affected [corrected]. Although ED screening might have a high yield, opportunistic screening is not feasible, with difficulties in staff engagement and patient follow up for diagnostic testing. Future studies might consider finger-prick fasting blood glucose through a patient's general practitioner for diagnosis.