OBJECTIVE: To develop a predictive model for the classification of seizure freedom under first-line monotherapy with levetiracetam in patients with newly diagnosed epilepsy. METHODS: Clinical data and routine EEG recordings of patients with newly diagnosed epilepsy who were started on a first-line monotherapy with levetiracetam were analyzed retrospectively. EEG had been acquired prior to the initiation of treatment in all patients. Patients who had experienced no further seizures until the last follow-up were labeled as seizure-free. Spectral EEG features (band power, peak power, and peak frequency) and functional connectivity were computed in each patient and, together with clinical data, formed the input features for the classification procedure. The BiMM-forest algorithm was used for classification and 5-fold cross-validation was conducted to evaluate the model performance. RESULTS: Seventy-four patients were analyzed (43 (58.1 %) female, 43/74 seizure-free. The mean classification accuracy was 75.5 % (95 %-CI = 46.1-92.4 %). The most predictive features for seizure-freedom were increased right frontal delta energy and delta peak energy as well as decreased central delta energy in the first routine EEG. SIGNIFICANCE: While significant above-chance predictions were not achieved in this study, a promising framework for the classification of treatment response on first-line monotherapy with levetiracetam based on pre-treatment EEG data alone was provided. Although negative, our results show trends that should encourage future, larger studies to develop EEG-based frameworks for the prediction of treatment response under specific anti-seizure medications.