Abstract: Rain will increase water content and reduce nutrition in the soil. Intelligent drip irrigation system is able to predict the beginning and ending of rain, and control the water-nutrition proportion in the irrigation system, thus saving water usage and building an optimum growing environment for crops. However, rain may disturb the communication between infrared transmitter and receiver, and cause errors in communications. A new error-detection-based infrared rain sensor was designed, which consists of a rain trap, a water tube, an infrared transmitter and receiver. The microcontroller can detect rain with the error in communication, while traditional rain sensors can only detect the beginning of rain, but not the ending. In addition, a method was proposed for optimal communication code selection, which was the selected code resulting in the largest error byte rate (EBR) when it was raining and the least EBR when it was not. The method showed that the code "1" had a significant effect on the EBR. Both theoretical analysis and experiments results proved that the optimal code was "255". Further experiments showed that the correction detection rate was larger than 0.9983 for continuous water flow, and between 0.7742 and 0.8307 for water drops. It is proved by experiment results that, the proposed sensor is better than the traditional rain sensors.