We investigate the mean square average consensus problem of multi?agent systems with directed topology being strongly connected. It is considered that the nonlinear multi?agent individuals are interfered by noise environment in the process of transmitting information, and a new control scheme is proposed to improve the control performance. Based on Lyapunov stability and It^o integral equation theory, the sufficient conditions for the mean square average consensus of the multi?agent systems are achieved. The allowable maximum delay interval with the same coupling strength is simultaneously obtained. Numerical simulations are also provided to demonstrate the effectiveness of theoretical analysis.