This paper presents a new generalized super-twisting algorithm (GSTA) controller with a timedelay estimator (TDE) for the tracking control of an autonomous underwater vehicle (AUV) under disturbances. For exploration tasks in confined areas, AUVs are typically equipped with sensors such as Doppler velocity logs (DVL) or acoustic modems to measure their position and speed. However, these sensors have a disadvantage in that their acquisition rates are low. This disadvantage, combined with the influence of external disturbances, directly influences position control because it requires partial knowledge of the system for proper operation. To solve this problem, the introduction of a TDE that estimates the vehicle hydrodynamics with delayed sensor information, and consequently, improves the controller performance is proposed. For controller validation, a stability analysis is presented, and two results are provided in the simulation. In the first test, the AUV is controlled to track a parameterized spiral-shaped path to validate the vehicle movement in the three directions of the plane (, ,). In the second test, the AUV performs path tracking in yaw and depth motions. In all numerical simulations, three types of disturbances/uncertainties are considered: (i) system parameters, (ii) sensor readings, and (iii) control.