Abstract: Continual spray has low efficiency and can cause large waste of pesticides. It can contaminate the soil and water environment. In contrast, targeted spray can reduce the droplets deposition on non-target areas to ensure the spray effect and to reduce pesticide. In order to solve these problems, a flexible targeted spray device has been developed. This device is composed of the flexible targeted spray unit and the target detection unit. The length of the device is 4.5 m, and the width is 1.5 m. The height of the spray support frame of this device can be adjusted during the range of 2.0-3.0 m. The flexible targeted spray unit consists of pesticide container support frame, a 400 L pesticide container, a diaphragm pump, a solenoid valve, a spray support frame, flexible pipes and nozzles group, which is mounted on the three-point hitch of DFH MS-304 tractor. The size (length, width and height) of support frame of pesticide container is 1.5 m × 0.8 m × 1.0 m. The support frame is divided into 2 layers. A 400 L horizontal pesticide container is placed on the upper and a diaphragm pump is placed on the lower. Diaphragm pump is connected with the solenoid valve, and is driven by the tractor PTO shaft. The solenoid valve is connected by the flexible pipes which are connected with the nozzles. To select the most appropriate spray method of the device, 4 spray methods are designed to test the droplet deposition rate. The first method is continual spray that the solenoid valve normally opens, i.e. the nozzles have been always in a state of spray. The second approach is targeted sprayⅠ by using ultrasonic distance measurement sensors close to the head sensor or the back sensor of the truck. The solenoid valve will open to spray when the relative position of spray unit and canopy is under the condition like that head sensor detects that there is a tree below the spray unit and the back sensor is unable to detect the target, and it will stop spraying under the condition like that head sensor is unable to detect the target and the back sensor detects that a tree is getting away from the area below the spray unit. The third method is targeted sprayⅡ and the fourth approach is targeted spray Ⅲ. Simulated target of fruit tree with 4 m plant spacing, 1.7 m height and 1.1 m canopy diameter is used as the test object. The experiment results show that corresponding droplet deposition rates of 4 spray methods are 40.3%, 50.4%, 77.8% and 86.0%, respectively. It can be seen that the droplet deposition rate of targeted sprayⅠis similar to continual spray, and it cannot significantly improve the droplet deposition rate. However, the droplet deposition rates of targeted sprayⅡ and Ⅲ are twice as large as continual spray's, and the droplet deposition rate is significantly increased. Therefore, targeted spray Ⅲ is chosen as the device's spray method. Target tree with 2.1 m canopy diameter is used as the test object to verify the chosen targeted spray method whether to increase the droplet deposition rate. The result is obtained that the droplet deposition rate is 88.4%. This experiment shows 2 laws about the droplets deposition. Firstly, the quantity of droplets deposition in the canopy is significantly higher than that among the trees, which reflects the characteristics of targeted spray. Secondly, the quantity of droplets deposition in the middle region of canopy is higher than that in edge region for the reason that the spray time of edge region is less than the middle region. This phenomenon is to meet the demand of fruit tree canopy for pesticide.