In recent works, well-known three-dimensional localization methods studied in the aerospace field have been revisited for applications on multirotor aerial vehicles (MAVs). However, most of these classic methods employ stochastic estimators that are asymptotically stable, in a stochastic sense, and exhibit high sensitivity to disturbances and model uncertainties. The goal of this paper is to propose and evaluate a novel solution to the localization problem of MAVs, employing multivariable robust observers based on sliding-mode techniques. Aiming to improve on the existing methods usually based on the extended Kalman filter, this paper investigates sliding-mode techniques to reach finite-time stabilization of the estimation error and provide robustness to disturbances and uncertainties. The super-twisting algorithm (STA) is considered as a starting point for its recognized performance when used to design differentiators. In particular, a modification of the STA is proposed, replacing one of its terms with a certain time-varying function that allows the upper bound of the settling time of the resulting algorithm to be a direct adjustable parameter. The proposed algorithm’s behavior is numerically evaluated and is shown to yield the predicted properties even in the presence of bounded disturbances and uncertainties. Additionally, an attitude determination problem employing the proposed algorithm is presented as an application. The three-dimensional attitude and angular velocity of an MAV are accurately estimated under strict settling-time restrictions, using only the vector measurements provided by an accelerometer and a magnetometer.