This paper considers mean-square bounded consensus(MSBC) problems of double-integrator multi-agent systems(MASs) with both additive system noises and communication noises. It is shown that systems can achieve MSBC when multiple additive noises coexist. Applying algebra, graph theory and random analysis, several necessary and sufficient conditions for MSBC of systems under fixed and switching topologies are respectively established. Additionally, the correctness of the obtained stochastic consensus results is verified by simulations.