The high moisture content of grain would result in quality deteriorations such as mold and mildew during storage and transportation. For effective detection of grain moisture, a forked-finger capacitive sensor in the form of triangular prism is proposed. The effect of the sensor structural parameters on its capacitance, electric field distribution, sensitivity and penetration depth are investigated using finite element simulations. The results show that an optimised effective penetration depth of 2.73 mm for grain moisture detection can be achieved when the finger width, number, length and spacing are 50 μm, 25.4 mm and 50 μm, respectively. This work may provide a new approach for the structural design of capacitive grain moisture sensors.