This paper addresses the attitude tracking problem of quadrotor unmanned aerial vehicles (UAVs) by introducing a novel predefined-time stable adaptive value iteration (PTS-AVI) control scheme. Unlike conventional value iteration-based adaptive dynamic programming (Ⅵ-ADP) methods—where the running cost is independent of the value function—the proposed cost function explicitly incorporates the value function $V$, thereby ensuring predefined-time stability (PTS) throughout both the training and deployment phases. This functional dependency was systematically addressed via an auxiliary time-scale partial differential equation (PDE) formulated in the $s$-domain. To circumvent the need to solve the complex Hamilton–Jacobi–Bellman (HJB) equation directly, a parameter update method was employed to approximate its solution, yielding an optimal control policy that satisfies predefined-time stability criteria. In addition, a high-precision predefined-time disturbance observer was designed to estimate and compensate for unknown disturbances. Both theoretical analysis and simulation results confirmed that the proposed control scheme guarantees state convergence to equilibrium within a user-specified time, regardless of initial conditions. This work differs from existing studies by integrating predefined-time stability requirements into the Ⅵ-ADP formulation and coupling it with a predefined-time disturbance observer for attitude tracking.