Arterial hypotension often signifies inadequate systemic perfusion. We hypothesize that in a heterogeneous emergency department (ED) population with clinically suspected circulatory shock, the severity of hypotension on presentation predicts in-hospital outcome. We performed a secondary analysis of patients with nontraumatic shock enrolled in a noninterventional, randomized, controlled trial. The setting was an urban, tertiary ED, census >100,000 visits per year. Patients included nontrauma ED patients, aged >17 years, with initial ED vital signs consistent with shock (systolic blood pressure <100 mmHg or shock index >1.0), and agreement of two independent observers for at least one sign and symptom of inadequate tissue perfusion. Measurements included interobserver agreement for signs and symptoms of shock, relationship between the depth and duration of ED hypotension and adverse hospital outcome (in-hospital mortality, need for intensive care unit services, and acute organ failure) and logistic regression analysis for independent predictors of adverse hospital outcome. Of 202 patients who qualified, 190 patients were included; the in-hospital mortality rate was 15%. The sign or symptom of shock with the highest interobserver agreement was "unresponsive" (kappa = 0.74). The adverse hospital outcomes increased with each decile decrease in the lowest ED systolic blood pressure (SBP) from 17% if SBP >89 mmHg versus 50% if SBP < 80 mmHg. Forty percent of patients with an adverse hospital outcome had sustained hypotension (all ED SBP <100 mmHg for > or =60 min). Sustained hypotension was the strongest independent predictor of an adverse hospital outcome (odds ratio 3.1; 95% CI 1.5-7.1). Mortality among patients who present to the ED with undifferentiated shock is high. The depth and duration of systolic blood pressure appears to have a dose-response relationship to adverse hospital outcome.