Acoustic fish driving technology, as an auxiliary fish passage measure and a non-physical barrier, is based on the use of sound signals to prevent or regulate fish behavior. The purpose of these techniques is to guide the fish away from dangerous areas, such as the water inlets of hydroelectric power stations, spillways, and ship locks, allowing them to easily locate the entrance to the fishway, which would help improve fish passage efficiency. Studying the negative phonotaxis behavior of fish is vital for establishing non-physical barriers using acoustic characteristics. However, there has been little research on verifying the effectiveness of acoustic fish deterrence technology in field environments. Therefore, this study used alternating sound playback to conduct negative phonotaxis experiments on grass carp (Ctenopharyngodon idellus) juveniles to explore their behavioral responses to different sounds. The experimental tank (3.6 m×1.1 m×1.0 m) was created in the waters of Xialao Creek in Yichang City, Hubei Province, with an average water depth of 0.5 m and an average flow rate of 0.06 m/s. The experiment used one single-frequency sound (1 000 Hz) and five complex sounds (fish swimming, engine, short-nosed crocodile call, pile driving, and yacht sounds), with a sound pressure level of (117.69±2.77) dB re 1 μPa. The effectiveness of acoustic fish-repellent technology has been proven, but there are only a few applications in practical engineering. On the one hand, the theoretical knowledge is not comprehensive, and on the other hand, there is a gap between theoretical research and practical engineering. Moreover, there are differences in proton movement (vibration) modes between indoor and natural environments. Compared with fish in an indoor environment, fish in natural waters tend to receive sound signals by proton movement rather than sound pressure. At the same time, the distribution of the sound field in natural and indoor environments also differs; thus, field experiments are necessary for the advancement of acoustic fish-repellent technology. Globally, studies on the negative phonotaxis of fish have mainly been conducted in vitro. Detailed studies using natural open water conditions are insufficient, and further field verification experiments are needed. Therefore, this study conducted experiments in natural open water, compared the sound field changes in the natural and indoor environments, and studied negative phonotaxis behavior by observing reaction time, initial reaction time, average reaction time, phonotaxis speed, movement time ratio, and other indicators. The results showed that when the complex sounds were played, the reaction times, tone trend speed and movement time ratio of grass carp were significantly higher than that of single tone and control group (P < 0.001), and the initial reaction time and average reaction time of grass carp were significantly lower than that of single tone and control group (P < 0.001). Among the complex sounds, the grass carp stimulated by the yacht sound had the largest response times and speed, while the grass carp stimulated by the fish swimming sound had the smallest response times and speed. In the complex sound, the first response time of grass carp stimulated by yacht sound was the shortest, which was (23.40±5.13) s. The first response time of grass carp stimulated by engine sound was (146.00±7.82) s, which was significantly lower than that of other complex sounds (P < 0.05). The average response time of grass carp stimulated by the sound of yacht and pile was (26.52±3.01) s and (28.76±4.07) s, respectively. The average response time of grass carp stimulated by fish swimming sound was (64.76±17.82) s. In the complex sound, the motion time ratio of grass carp stimulated by fish swimming sound was the highest, which was (98.47±0.48)%. The motion time ratio of grass carp stimulated by engine sound was (94.58±0.54)%. There were no significant differences in reaction times, initial reaction time, average reaction time and exercise time ratio between grass carp and control group when playing single frequency tone (P > 0.05). The experimental results indicated that the five complex sounds used in this study (fish swimming, engine, short-nosed crocodile call, pile driving, and yacht sounds) all had a deterrent effect on grass carp juveniles. This study not only enriches current knowledge of the negative phonotaxis behavior of fish but also provides a scientific basis for the design and optimization of sound-based fish deterrent facilities in practical engineering.