Previous research on alcohol dependent patients has shown that variations in eyeblink startle response can be used as an indicator of their emotional responses to alcohol-related stimuli. Postulating that reactions on substance associated stimuli are controlled by either a negative or a positive affective processing system, we expect that abstinent alcoholics react differently (within-group) in the emotional evaluation of alcohol cues. Furthermore, we assumed the startle response to covary with medication response to acamprosate and naltrexone. We measured 74 detoxified inpatients' acoustic startle responses while they were being presented with alcohol-related images as well as affectively negative, neutral, and positive pictures before they were randomized to pharmacotherapy. Group-mean startle peak amplitudes were lowest for alcohol-related cues. The relative startle response (alcohol cues set in relation to the other stimulus categories) did not correlate with craving for alcohol (OCDS) or alcohol cue induced self-ratings of arousal, valence and craving. Patients with a lower percentage of abstinent days in the 90 days prior to the last drinking day showed a lower ("more appetitive") startle response to alcohol cues. A survival analysis using the time to first heavy drinking day as the survival criterion revealed a significant interaction between alcohol-cue startle responses and medication type. The results indicate that the psycho-physiological measure of emotional evaluation of alcohol cues includes unconscious processing not reflected by conscious self-ratings. Furthermore, our result of a differential medication effect may encourage further studies to use biological characteristics to stratify patients as a step towards individualized treatment for alcohol dependence.