There is a fast-growing need for maintenance of railway track systems due to increasing the number of passengers and freight transport in the past few years. In this paper, a novel data-driven method is proposed to detect different maintenance needs of railway track systems using vertical acceleration data collected from an operational passenger train. The framework contains four modules. Firstly, a data pre-processing and cleansing is performed to extract useful data from the whole dataset. Then, condition-sensitive features are extracted from the raw data. In the third module, the best subset of measurement features that characterize the state of the tracks are selected using Analysis of Variance (ANOVA) algorithm which eliminates irrelevant characteristics from the time domain feature set of responses. Finally, a multilabel classification algorithm based on the Support-Vector Machine (SVM) is used to classify the type of maintenance needs of track. An open-access dataset from a field study in the United States is used in this study for validation of the proposed method. Results show that when an SVM is used, a higher accuracy can be achieved for detecting tamping and surfacing needs in comparison with k-Nearest Neighbors (KNN) and Decision Tree (DT). In addition, the impact of using different features selection methods, different classification algorithms and different accelerometer types (uniaxial and triaxial accelerometers) on the accuracy of the proposed approach is studied.