Objective Traditionally, drop-out has been investigated subsequent to treatment entry; yet some individuals "exit early," attending assessment but failing to commence the treatment offered. Early exit burdens administrative and clinical resources and means that individuals may not receive the care needed for recovery. This study aimed to describe the prevalence and characteristics associated with early exit at a statewide, outpatient eating disorder service. Method From a pool of 972 consecutive referrals of adults and youth (16+ years), two groups were formed; an "early exit" group of individuals who attended assessment but chose to exit the service prior to treatment entry and a "non-early exit" group that attended assessment and entered treatment. The groups were compared on sociodemographic, clinical, and administrative features. Results The prevalence of early exit was 18.7%. The early exit group exhibited less pathology compared with the non-early exit group; specifically, a lower presence of Axis I comorbidity (p =.04) and self-induced vomiting (p =.04). Discussion The findings, considered in the context of previous research, suggest that there are no measured features to date that are robustly associated with early exit from outpatient eating disorder services. Future research should investigate decision-making processes at assessment, to inform patient-centered approaches that optimize transition to treatment. © 2013 by Wiley Periodicals, Inc. (Int J Eat Disord 2013; 46:553-559) Copyright © 2012 Wiley Periodicals, Inc., A Wiley Company.