Soft robotic grippers can gently grasp and maneuver objects. However, they are difficult to model and control due to their highly deformable fingers and complex integration with robotic systems. This paper investigates the design requirements as well as the grasping capabilities and performance of a soft gripper system based on fluidic prestressed composite (FPC) fingers. An analytical model is constructed as follows: each finger is modeled using the chained composite model (CCM); strain energy and work done by pressure and loads are computed using polynomials with unknown coefficients; net energy is minimized using the Rayleigh–Ritz method to calculate the deflected equilibrium shapes of the finger as a function of pressure and loads; and coordinate transformation and gripper geometries are combined to analyze the grasping performance. The effects of prestrain, integration angle, and finger overlap on the grasping performance are examined through a parametric study. We also analyze gripping performance for cuboidal and spherical objects and show how the grasping force can be controlled by varying fluidic pressure. The quasi-static responses of fabricated actuators are measured under pressures and loads. It is shown that the actuators’ modeled responses agree with the experimental results. This work provides a framework for the theoretical analysis of soft robotic grippers and the methods presented can be extended to model grippers with different types of actuation.