Fused deposition modeling (FDM), one of representative additive manufacturing (AM) technologies, has been widely used for fabricating functional parts with geometrical complexity. However, it has suffered from degraded part quality and low process reliability and controllability. Therefore, it is of much significance to develop a monitoring and diagnosis system for the FDM process to overcome such drawbacks. In this paper, a data-driven FDM process monitoring and diagnosis system is developed by using two types of sensors – an accelerometer and an acoustic emission (AE) sensor. A large number of experimental data, collected from the accelerometers and AE sensor under healthy and faulty process states, are processed to obtain a critical feature – a root mean square (RMS). The RMS values are then used for training the FDM process monitoring and diagnosis models based on a support vector machine (SVM) algorithm and a k-fold cross validation approach. In particular, the SVM-based models for the odd- and evennumbered layers of one FDM specimen are developed. For a real-time validation in a factory floor, the non-linear SVM-based models using the acceleration signals are used for the software development. The diagnosis accuracy is better than 87.5%, and an applicability of the models is verified.