The development of unmanned aerial vehicle (UAV) helicopters is required to solve complex engineering problems and it contains a number of research efforts from multiple disciplines. This thesis aims to present a systematic and comprehensive methodology of the development of UAV helicopters. The methodology can provide the researchers with necessary knowledge to advance and validate their work, and remove the possibility of information loss due to the knowledge integration of multiple disciplines. To achieve this goal, research is conducted as follows, The development of UAV testbeds is to validate and test flight controllers by carrying out actual flight experiments. The onboard hardware components are carefully selected and integrated with the outboard processors and the motion capture system. The software system is developed in a timely manner and its functionality is enriched by incorporating the capability of flexible reconfiguration. To acquire high fidelity and accurate sensor data, data processing algorithms are developed. An accurate dynamic model of the UAV helicopter can benefit the control law design. The nonlinear dynamic model is derived from the first-principles modeling. To improve the accuracy and enlarge the flight envelopes, the derivation of mathematical equations has additionally taken into account the aerodynamic and proximity effects. The derived models cover the helicopters dynamics not only in hover, but also in translational flights, particularly at high speed. Several control laws have been designed for the UAV helicopters with different control purposes. PID control is applied to realize some basic flights of the vehicles. To achieve high-performance control capability for large flight regimes, advanced and sophisticated control laws have been employed. The closed-loop state feedback control has been employed in the design. Reduced-order observer is used to estimate the unmeasurable states. An augmented integral model is developed to eliminate the steady-state errors. The control and observer gains are optimized with linear quadratic regulator (LQR) methods. The problems resulting from the unmodeled dynamics and external disturbances are addressed by the integration of sliding mode control approach in the overall control law design. The results from the simulation and flight experiments are used for analysis and design of the control algorithms.