A wealth of functional magnetic resonance imaging monetary incentive delay task (MIDT) research has shown alcohol dependency is associated with a hypoactive striatal response during gain anticipation (gain > neutral) and loss anticipation (loss > neutral). Electroencephalography (EEG) holds clinical advantages over fMRI (high temporal resolution, low cost, portable) however its use to study reward processing in alcohol dependence is limited. We aimed to carry out the first EEG MIDT (eMIDT) study in alcohol dependence. 21 abstinent alcohol dependent individuals and 26 controls performed an MIDT while neural activity was recorded using 64-channel EEG. Trial averaged event-related potentials (ERPs) and single-trial machine learning discriminant analyses were applied to EEG data. Clinical variables related to severity of dependence were collected and relationships with ERP data explored. Alcohol dependent individuals, compared with healthy controls, had blunted cue-P3 amplitudes for gain and loss anticipation (interaction: p = 0.019); and elevated contingent negative variation amplitudes for all conditions (gain, loss, neutral)(main effect: p < 0.001) which was associated with increased alcohol consumption (p = 0.002). The machine learning analyses demonstrated alcohol dependent individuals had reduced ability to discriminate between loss and neutral cues between 328 – 350 ms (p = 0.040), 354 – 367 ms (p = 0.047) and 525 – 572 ms (p = 0.022). The eMIDT approach is demonstrated to be a low-cost, sensitive measure of dysfunctional anticipatory reward processing in alcohol dependence, which we propose is ideal for big data approaches to prognostic psychiatry and translation into clinical practice.