To achieve variable rate spraying, a pressure-based variable rate spaying system was designed to control the spray rate by adjusting the opening percentage of electrical regulating valve at the bypass pipeline. This returns excessive flow discharged by a piston diaphragm pump according to the flow rate of the spray pipeline, and adjusts the boom width by shutting ON or OFF selected solenoid valves at each section. The ARM9 based variable rate controller was designed to measure the system pressure and flow rate, and generate the control signal. A commercial 3W-250 boom sprayer with 12 flat fan spray nozzles was modified to a variable rate sprayer and mounted behind the LOVAL TA800 tractor. The 12 nozzles were divided to 6 boom sections and the pump was derived by the power take-off shaft of the tractor. The variable rate sprayer was represented by a directed graph of fluid network that consists of a set of junctions, called nodes, and certain lines joining a pair of nodes, called the edges. To reduce the complexity of the spray network, the fluid resistance of the short pipeline was ignored, some nodes were merged, one virtual node was increased to represent the external atmospheric pressure, and finally the graph of the variable rate spraying system involved 6 nodes and 12 edges. The flow rate and pressure distribution within the spray network under, steady state conditions was described by the junction of continuity equations and the loop energy equations. The electrical regulating actuator was also modelled to describe the relationship between the voltage control signal and the valve opening percentage. The fuzzy-PID control algorithm was adopted for the nonlinear, time-varying variable rate spraying system to achieve better performance than with the conventional PID control algorithm. The fuzzy control algorithm was used for tuning the PID parameters online. The Smith predictor based on the system model was introduced to overcome the side effects of long time-lag that included the response delay of the flow meter at a low spray rate and the electrical regulating valve and to stabilize the delay of the spray rate. The variable rate spraying experiments were accomplished at 3 different desired nozzle spray rates which are 0.864, 0.72 and 0.576 m3/h for the Smith-Fuzzy PID control and the Fuzzy PID control algorithm without boom section control while the tractor was running at an idle speed of approximately 1 m/s. The results showed that the Smith-Fuzzy PID control algorithm has the better dynamic response and the higher application accuracy than the PID control algorithm, the overshoot and the steady mean absolute error of the former control algorithm were less than 13.1% and 3.52% respectively, and both decreased with the raise of the desired spray rate. The boom section control experiments were also conducted for the Smith-Fuzzy PID control and the PID control algorithm. Two boom sections were turned off at the same time while the spray rate stabilized at a desired spray rate for each nozzle is 0.06 m3/h. The results revealed that the dynamic response curve of the former algorithm had a shorter response time and no overshoot after the two boom sections shut off, and the mean absolute error was 0.017 m3/h and 0.0185 m3/h respectively.