Instrumented crutches are useful for many rehabilitation tasks, including monitoring the correctness of crutch use, analyzing gait properties for patients with lower-limb impairments, as well as providing sensory data for controlling lower-body robotic orthoses. In this paper, we describe the design and analysis of an instrumented crutch system equipped with low-cost accelerometer and pressure sensors to estimate all components of the ground reaction force (GRF), providing a well-defined and physically meaningful sensory output for practical applications. We propose an angle-dependent quadratic model to map pressure and inclination data to force components, which we identify using least-squares methods. Through systematic characterization experiments, we first show that our model can predict GRF vectors with less than 7% rms errors in all axes for fixed crutch angles used for training. Subsequently, we generalize the model to crutch angles other than those used for training, showing that rms estimation errors remain below 7% for all axes. Finally, we assess measurement accuracy and performance under dynamic loading conditions with time-varying crutch angles, showing that errors still remain below 8% under realistic conditions.