The repeated game literature studies long run/repeated interactions, aiming to understand how repetition may foster cooperation. Conditioning future behavior on past play is crucial in this endeavor. For most situations of interest a given player does not directly observe the actions chosen by other players and must rely on noisy signals he receives about those actions. This is typically incorporated into models by defining a monitoring structure, that is, a collection of probability distributions over the signals each player receives (one distribution for each action profile players may play). Although this is simply meant to capture the fact that players don.t directly observe the actions chosen by others, constructed equilibria often depend on players precisely knowing the distributions, somewhat unrealistic in most problems of interest. This paper aims to show the fragility of belief free equilibrium constructions when one adds shocks to the monitoring structure in repeated games.