Abstract: The global asymptotic stability of cellular neural networks with uncertain parameters and time-varying delays was investigated, thus giving a new criterion for stability, where the relevant uncertainties are bounded and the uncertainties of system parameters are in relation to time hence restricted in a certain range. The time delay function changes with time, and its derivative is less than 1. For such a complicated problem, there is no highly mature theory about it internationally until now. So, the model to be developed relevantly will be more general in comparison with that as shown in earlier works. Framing properly a new Lyapunov funtion and taking LMI (linear matrix inequality), the sufficient conditions are given to system stability. Two simulative examples are given to illustrate theoretically the effectiveness of the method proposed.