We propose a wrapped statistics-based approach for phase estimation from noisy reconstructed interference fringes in digital holographic interferometry. The state space model required here is formed by Taylor series expansion of the phase function as state model and the wrapped dynamical system as measurement model. Prediction of the state using Kalman filter is straightforward since the state model is linear. However, the non-linearity issue induced due to the wrapping of the measurements is handled by changing the innovation correction step, which accounts for the probability of wrappings. Through the simulation and experimental study, we have shown that the proposed approach is robust to both, noise in fringe pattern as well as the dynamic range of the phase pattern, simultaneously. Moreover, it outperforms when compared with the other state-of-the-art phase retrieval approaches. © 2016 Taylor & Francis.