Cable-Driven Parallel Robots (CDPR) employ extendable cables to control the pose of an end effector (EE). If the number of cables is smaller than the degrees of freedom of the EE, and cables have no special arrangement reducing the EE freedoms, the robot is underactuated, and the EE is underconstrained: as a consequence, the EE preserves some freedoms even when all actuators are locked, which may lead to undesirable free motions. This paper proposes a novel methodology for the identification of the EE inertial parameters of these robots. Inertial parameters are useful, for example, in the application of feedforward control techniques. The main merit of our approach is that it does not require force or torque measurements, and only a subset of the robot kinematic variables needs to be measured. The method consists in the application of the EE internal-dynamics equations along a free-motion trajectory, also referred to as self-motion zero dynamics. This results in an over-determined system of equations that are linear in the EE inertial parameters (the Free-motion Internal-Dynamics Identification Model, FIDIM); the said system is solved according to the Total-Least-Square technique. Free-motion trajectories that are optimal for identification purposes are investigated and experimentally tested on a 4-cable robot. FIDIM is then applied, statistical analysis is performed, and the experimental results are cross-validated against additional free-motion trajectories.