In this paper, the observer-based state estimation of discrete-time Takagi-Sugeno fuzzy systems is enhanced markedly by developing a distinctive multi-instant gain-scheduling law. Firstly, the updated information of both the past-time and the current-time normalized fuzzy weighting functions is fused into the proposed multi-instant gain-scheduling law in a comprehensive way, and it is important to point out that those existing ones reported in the literature belong to special cases of ours. Secondly, a set of time-varying free matrices for constructing more relaxed fuzzy observers are introduced for the first time in accordance with different operating modes. Moreover, the time-varying free matrices are more general than those previous time-invariant ones and most of them no longer need to be positive or negative definite. Therefore, a significant reduction in conservatism can be achieved at the expense of a slight increase in offline computing burden over those competing approaches. Finally, a set of numerical simulations are implemented to verify the superiority of the designed approach.