In electric power systems, detecting inter-area oscillations is crucial to the system operators for maintaining the security of the grid – especially in the case of unstable oscillatory behaviour. However, extracting information from unstable, noisy, signals is complicated with conventional signal processing tools suffering from insufficient adaptability. In this paper, we propose a method based on Empirical Wavelet Transform (EWT) to estimate in real-time the dominant inter-area modes in electricity grids. EWT extracts the inherent modulation information by decomposing the signal into its mono components under an orthogonal basis. The instantaneous amplitude and instantaneous frequency is estimated by applying Hilbert transform from the narrow band components of the decomposed EWT signal. The performance of the proposed method is demonstrated using the Nordic test system.