Automatic multi class text classification is a machine learning task which categorizes document to one among a predefined set of classes. In recent years, deep learning technique such as Recurrent Neural Networks (RNNs) has become state-of-the-art model for a variety of machine learning problems. This paper introduces the scope of Long Short Term Memory (LSTM) - a type of RNN, for multi class text classification. LSTMs are capable of learning long-term dependencies while avoiding the vanishing gradient problem usually found in neural network algorithms. The proposed system is carried out in Reuters corpus, a dataset of 11,228 news wires from Reuters, labeled over 46 topics.