This paper gives a new sufficient condition for Cellular Neural Networks with Delay (DC-NN's) to be completely stable. A DCNN is said to be completely stable if and only if its trajectory converges to and equilibrium point for any initial condition. Most of the conventional stability criteria were obtained by constructing Lyapunov functions. On the other hand, a convergence theorem of the Gauss-Seidel method, which is a well-known iterative technique for solving linear algebraic equations, plays an important role in this paper.