Inter-area oscillations in power systems needs to be monitored and accurate estimation of frequency and damping of the oscillations in real time is vital as they provide considerable insight into system stability. Zolotarev polynomial based filter bank (ZPBFB) is proposed for decomposing the PMU signals into monocomponents. Modal frequency and damping are obtained from the decomposed monocomponent signals through eigen realization algorithm (ERA). The salient property of ZPBFB lies in signal decomposition with sum of decomposed signals produced by filter bank, without oversampling, is an exact replica of input signal with delay. A narrow bandwidth of 0.1 Hz is achieved in the frequency range of 0 to 1 Hz with ZPBFB. The robustness of the proposed method to noise is demonstrated using Monte Carlo simulations. The proposed ZPBFB and cosine modulated filter bank (CMFB) are implemented on two-area test system and 16-machine, 68-bus system and the performances are compared. The efficacy of the proposed method is also demonstrated on three real time signals recorded by wide area frequency measurement system (WAFMS) at 3 locations in India on November 30, 2011. © 1969-2012 IEEE.