Attending physician judgment is the traditional standard of care for emergency department (ED) admission decisions. The extent to which variability in admission decisions affect cost and quality is not well understood. We sought to determine the impact of variability in admission decisions on cost and quality. We performed a retrospective observational study of patients presenting to a university-affiliated, urban ED from October 1, 2007, through September 30, 2008. The main outcome measures were admission rate, fiscal indicators (Medicaid-denied payment days), and quality indicators (15- and 30-day ED returns; delayed hospital admissions). We asked each Attending to estimate their inpatient admission rate and correlated their personal assessment with actual admission rates. Admission rates, even after adjusting for known confounders, were highly variable (15.2%-32.0%) and correlated with Medicaid denied-payment day rates (p=0.038). There was no correlation with quality outcome measures (30-day ED return or delayed hospital admission). There was no significant correlation between actual and self-described admission rate; the range of mis-estimation was 0% to 117%. Emergency medicine attending admission rates at this institution are highly variable, unexplained by known confounding variables, and unrelated to quality of care, as measured by 30-day ED return or delayed hospital admission. Admission optimization represents an important untapped potential for cost reduction through avoidable hospitalizations, with no apparent adverse effects on quality.