<p>This article addresses the issue of Lagrange stability of quaternion-valued memristive neural networks. First, by scale-limited Halanay inequality, new Lagrange stability algebraic conditions are obtained. Considering that the memristive connections are switching between two different parameters, thus, the memristive model is equivalent to a robust system by introducing some new matrices, we then use the generalized matrix measure to solve the time scales matrix norm issues, and all the derived scale-limited sufficient criteria not only apply to continuous-time system and their discrete-time analogs, but also satisfy the system with uncertain terms. Eventually, illustrations are addressed to reflect the solvability and practicability of the strategy.</p>