The tillage depth had a significant influence on root growth of crops, energy consumption of agricultural implements and tillage quality in soil tillage. It was extremely difficult to measure tillage depth accurately in real time because of straw mulching and roughness in the field as well as the vibration of tractor. A tillage depth detecting device based on Kalman filter and data fusion was designed in order to improve the accuracy of tillage depth detecting in real time. An ultrasonic sensor and an infrared sensor detecting tillage depth were used by time of flight and triangulation measurement, respectively. The data from two sensors were filtered and fused through Kalamn filter and data fusion algorithm. Besides, capacitance touch screen and memory chip were added in this device for inputting and outputting data expediently and further research. The detecting results of the infrared sensor were superior to that of ultrasonic sensor in even ground and the results were contrary in straw mulching ground. The data filtered and fused by Kalman filter could make full advantage of the useful data of two sensors in varied environments. The averages of ultrasonic sensor filtered data were 29.51cm and 38.79cm, the coefficients of variations of tillage depth were 2.51% and 3.10% in field experiment when the standard tillage depth were 30cm and 40cm. The averages of infrared sensor filtered data were 32.06cm and 41.52cm, the coefficients of variation of tillage depth were 2.41% and 2.76% under the same conditions. The averages of Kalman filtered and fused data were 30.06cm and 39.95cm, the coefficients of variation of tillage depth were 1.07% and 1.00%. The device used filtered and fused detecting data had better performances in detecting tillage depth accurately and its trend than that of single sensor.