This paper is concerned with the robust stochastic stability analysis for two-dimensional (2-D) discrete state-multiplicative noisy systems (SMNSs) in the Roesser form. We first derive a new sufficient condition under which linear discrete 2-D SMNSs are 2-D robustly stochastically stable. The underlying problem can then be recast as a convex problem expressed by linear matrix inequalities, which can be facilitated using existing numerical algorithms. We then apply the obtained result to examine the 2-D robust stochastic stability for 2-D digital filters in the Roesser form with random coefficient variation and saturation overflow arithmetic based on free-weighting matrices and diagonally dominant matrices. An illustrative example is presented to verify the usefulness and potential of the proposed results. (C) 2016 Elsevier Ltd. All rights reserved.