The food industry is dependent on human labor for tasks that require tactile sensing, due in part to a lack of robotic sensors that are delicate enough to interact with food. In this paper we present the design, modeling, and performance of a durometer constructed with soft materials. We performed experiments to investigate the sensor’s material selection, repeatability, drift, probing speed, and calibration. We also integrated the sensor into a commercial soft robotic gripper and used it to measure the hardness of an orange. The orange would be damaged by a traditional durometer, but the soft durometer left no visible marks. These results suggest that soft robotic sensors can benefit the food industry and overcome limitations associated with the interaction between robotic systems and fragile objects.