The author studies the stochastic Itô system \[ dx=A(t)x dt+\sum_{i=1}^{m}B_i(t)x dW_i(t).\tag{1} \] Here, \(A(t),B_1(t),\ldots,B_m(t)\) are deterministic matrices defined on the semi-axis \(t\geq 0\) and \(W_1(t),\ldots W_m(t)\) are totally independent scalar Wiener processes determined on a complete probability space \((\Omega,F,\text{P})\). These conditions provide the existence and uniqueness of a strong solution \(x(t,x_0)\) satisfying the initial condition \(x(0,x_0)=x_0\). Generalizing the well known result from the theory of ODEs [see \textit{Yu. A. Mitropol'skij, A. M. Samojlenko} and \textit{V. L. Kulik}, Studies in dichotomy of linear systems of differential equations by means of Lyapunov functions. Kiev: Naukova dumka (1990; Zbl 0776.34041)], the author proves that system (1) is exponentially dichotomic in mean square once there exists a symmetric bounded and differentiable matrix \(S(t)\) on \([0,\infty)\) satisfying the condition \[ \dot S(t)+A^T(t)S(t)+S(t)A(t)+ \sum_{i=1}^{m}B_i(t)S(t)B_i(t)0\), where the random variable \(Q(\omega)\) is bounded with probability 1.