This study proposes a fault diagnosis approach for electric motors using Radial Basis Function (RBF) neural networks. The proposed method utilizes features extracted from both time and frequency domains to improve classification performance. These features are directly extracted from vibration signal segments, which preserves essential fault-related signatures in both time and frequency domains. This approach enhances diagnostic accuracy by avoiding the loss of critical information during excessive pre-processing, while maintaining computational efficiency through the use of simple yet informative statistical and spectral descriptors. Three bearing conditions including healthy, inner ring fault, and outer ring fault are considered. The RBF neural network is trained using a subset of experimental data corresponding to known machine conditions. The trained model is then evaluated using the remaining data to assess its classification accuracy and generalization capability. The results indicate that combining time- and frequency-domain features significantly improves fault discrimination. Moreover, the use of RBF networks provides a reliable solution with low computational burden, making the proposed method suitable for practical fault detection systems in industrial applications.