We present, in this article, an analytical study of the phase reconstruction error for MCAO systems. Two approaches are considered; a classical estimator based on a Least Square (LS) approach and a Maximum A Posteriori (MAP) estimator which uses the prior knowledge we have both on the measurement noise and the turbulence volume statistics. The effects of these modes both on the phase reconstruction and on the phase correction error are studied and quantified. T is shown that, using a MAP approach, a large part of the unseen modes can be extrapolated using correlations between unseen eigenmode coefficients and well-measured coefficients. We observe that the use of a MAP estimator allows a significant gain in terms of correction quality in the whole field of view.© (2001) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.