This brief is concerned with the problem of learning observer-based robust H_∞ fault-tolerant Control (FTC) for Takagi-Sugeno (T-S) fuzzy descriptor systems with time-delay affected by both actuator faults and external disturbances. The proposed observer allows simultaneous estimation of system states and actuator faults. Based on the state estimation and the actuator fault reconstruction, a FTC is designed to maintain the performance of the faulty system. Using the H optimization technique, the analysis and design conditions of the fuzzy learning observer-based FTC are provided and then formulated into a set of delay-dependent linear matrix inequalities (LMIs) which can be solved in a single step. An example is finally presented to validate our findings.