A critical issue of a Cognitive Radio Network is the interference generated by secondary users simultaneously accessing and transmitting over the primary user spectrum band. In this paper, we introduce and study a distributed and opportunistic access scheme for MIMO ad-hoc cognitive radio networks identified as OPTIM-COG (OPporTunistic Interference control for Mimo COGnitive radio). OPTIM-COG is based on simple power stimuli issued by PU transmitters. These stimuli are basic control messages exchanged, at a known power, by both primary transmitter and primary intended receiver for their internal power control. A periodic repetition of this power stimulus and the power control generated in the primary network are used by secondary nodes to get their transmission opportunities. These opportunities are determined with a twofold goal: on the one hand the secondary node will transmit to its secondary receiver only if the quality of the already established primary connection does not decrease below a minimum level, on the other hand the resulting opportunistic access exploits the MIMO performance improvements. We describe the fully distributed version of this access scheme and we theoretically demonstrate, via a game theoretical formulation, that the resulting power allocation is the unique Nash Equilibrium. Performance results present secondary user access maps in the 2D plan, where both MIMO benefits as well as the presence of multiple active secondary links can be represented. Both the cases of a perfect interference measurement and imperfect channel estimation are evaluated.