Alcohol-related brain damage (ARBD) is a common and prognostically relevant complication of alcohol dependence. Clinical sensitivity for ARBD by routine mental state examination (MSE) is important, but unlikely given the documented insensitivity of the MSE in detecting less subtle disorders. We sought to determine the predictive capacity of a well-documented MSE in 32 detoxified alcohol-dependent patients who had undergone neuropsychological testing. Seventeen patients were found to have impaired performance on neuropsychological testing. The initial medical diagnoses and short-term memory tests did not correlate with neuropsychological findings. Ratings of the non-memory component of the MSE were a weak predictor of neuropsychological impairment. In the absence of any adequate 'screening' procedure, we conclude that all alcohol-dependent patients should be considered at high risk for ARBD and require full investigation.