Summary: We consider the parabolic equation with random coefficients: \[ D_ tu^{\epsilon}(t,x)=\sum_{ij}a_{ij}(t\epsilon^{- 1},x,\omega)D_{x_ ix_ j}u^{\epsilon}(t,x)+\sum_{i}b_ i(t\epsilon^{-1},x,\omega)D_{x_ i}u^{\epsilon}(t,x). \] We show that \(u^{\epsilon}(t,x)\) converges to the solution \(u^ 0(t,x)\) of the averaging equation: \[ D_ tu^ 0(t,x)=\sum_{ij}E(a_{ij}(t\epsilon^{-1},x,\omega))D_{x_ ix_ j}u^ 0(t,x)+\sum_{i}E(b_ i(t\epsilon^{-1},x,\omega))D_{x_ i}u^ 0(t,x). \] Also, the fluctuation process \(y^{\epsilon}(t,x)(\equiv (u^{\epsilon}(t,x)-u^ 0(t,x))/\sqrt{\epsilon})\) converges weakly to a generalized Ornstein- Uhlenbeck process on \({\mathcal S}'\).