Major Depressive Disorder (MDD) is a prevalent mental health condition with limited treatment efficacy. Electroencephalography (EEG) offers a promising avenue for exploring the neural underpinnings of MDD. However, existing EEG-based approaches often rely on time-consuming feature extraction methods and require a large number of recording electrodes. This can be impractical for clinical settings. This study proposes a novel approach using Long Short-Term Memory (LSTM) networks to classify MDD based on 128-channel EEG resting state recordings. The dataset used in this research is MODMA (Multi-model Open Dataset for Mental-disorder Analysis). The LSTM used in the research is a type of recurrent neural network, that can directly learn from raw EEG data, eliminating the need for manual feature engineering. The results demonstrate an accuracy of 72% in distinguishing individuals with MDD from healthy controls. This research suggests the potential of LSTM for accurate and efficient MDD classification using EEG data, paving the way for improved diagnosis and treatment.