Clinical and medico-legal decisions often require knowledge of alcohol impairment that is not necessarily revealed by an individual's appearance, and in turn, may not necessarily reflect level of blood alcohol. This study compares clinical signs and symptoms with measured and estimated blood alcohol concentrations (BACs). Method: Individuals (n = 384) perceived to be under the influence of alcohol at presentation to an emergency department were assessed by physicians and nurses for clinical features of alcohol intoxication (alcohol symptom checklist, ASC), who were asked to estimate the patient's BAC. Relation to measured BACs was assessed by correlation. Results: BACs ranged from 0 to 418 mg/100 ml. The correlation between the estimated BAC and measured BAC was r = 0.513. Measured BAC correlated with ASC r = 0.250. In subjects without a history of chronic drinking (n = 134) there was a better (P < 0.05) correlation with the ASC score (r = 0.363) versus measured BAC compared with that for chronic drinkers (r = 0.154). The positive predictive value of estimating BAC at or above a particular BAC cut-off decreased from 93.2% at 100 mg/100 ml to 37.7% at 300 mg/100 ml (P < 0.05). Conclusions: Measured BAC does not correlate well with the outward physical signs of intoxication, especially for chronic drinkers. There is a need for further education on how tolerance masks clinical signs of intoxication for the chronic drinker. BACs should be measured especially in the obtunded where no history (symptoms) can be given by the patient.