In this paper, the performance augmentation of underactuated fingers through additional actuators is presented and discussed. Underactuated, also known as self-adaptive, fingers typically only rely on a single actuator for a given number of output degrees of freedom (DOF), generally equal to the number of phalanges. Therefore, once the finger is mechanically designed and built, little can be done using control algorithms to change the behavior of this finger, both during the closing motion and the grasp. We propose to use more than one actuator to drive underactuated fingers to improve the typical metrics used to measure their grasp performances (such as stiffness and stability). In order to quantify these improvements, two different scenarios are presented and discussed. The first one analyzes the impact of adding actuators along the transmission linkage of a classical architecture while the second focuses on a finger with a dual-drive actuation system for which both actuators are located inside the palm. A general kinetostatic analysis is first carried out and adapted to cover the case of underactuated fingers using more than one actuator. Typical performance indices are subsequently presented and optimizations are performed to compare the best designs achievable with respect to stiffness and grasp stability, depending on the number of actuators.