Abstract. The 2D absolute phase estimation problem, in interferometric applications, is to infer absolute phase (not simply modulo-2π) from incomplete, noisy, and modulo-2π image observations. This is known to be a hard problem as the observation mechanism is nonlinear. In this paper we adopt the Bayesian approach. The observation density is 2π-periodic and accounts for the observation noise; the a priori probability of the absolute phase is modeled by a first order noncausal Gauss Markov random field (GMRF) tailored to smooth absolute phase images. We propose an iterative scheme for the computation of the maximum a posteriori probability (MAP) estimate. Each iteration embodies a discrete optimization step (Z-step), implemented by network programming techniques, and an iterative conditional modes (ICM) step (π-step). Accordingly, we name the algorithm ZπM, where letter M stands for maximization. A set of experimental results, comparing the proposed algorithm with other techniques, illustrates the effectiveness of the proposed method.