This dissertation presents measurement-based system identification methods that help to improve the reliability and operational stability of modern power systems. One of the important factors in maintaining stability is to know the damping and frequency of the oscillatory modes for all system operating conditions. Widespread use of synchrophasors has paved the way for several measurement-based approaches for estimating oscillation modes, damping, and frequency. These methods provide more accurate results than model-based approaches. This dissertation studies the effectiveness of state-of-the-art mode-estimation and proposes a framework based on subspace identification that provides a more accurate modal estimation in real-time. The proposed framework can also classify the oscillation types overserved in the measurements. Oscillatory modes are not observable at all measurement locations. Towards this, in this work, an optimal signal selection method is proposed based on subspace affinity. This helps to reduce the computational time of the modal estimation algorithms, which is critical for any real-time monitoring tool. This work also proposes approaches for mitigating oscillations. First, a method for locating the source of oscillations using the energy of oscillations is presented. Second, a framework for updating power system models based on measurements is proposed that helps system operations and planning. Finally, an integrated control framework for a wide-area damping controller (WADC) is proposed which mitigates different types of oscillations observed in the system. Effectiveness of the overall framework is tested with IEEE test systems and with real-life models with relevant data-sets. The studies show that the proposed approaches can improve the system’s situational awareness.