Small and Medium size Wind Turbines (SMWTs) can be used for the production of electricity in residential areas, or integrated into hybrid systems, which makes them more attractive in remote areas with difficult access to electricity from the grid. However, many of the SMWTs are installed in locations with low wind speeds, reducing remarkably their annual energy output. Furthermore, SMWTs are typically operated in sub-optimal conditions because they are not completely understood in the real environment. In a design stage, typically a wind tunnel is used to determine the aerodynamic efficiency, but latest research suggests that a turbine in the field behaves differently, especially when the wind is unsteady. Therefore, the aerodynamic performance of a wind turbine tends to be different in the field. Unfortunately, field testing is expensive, and requires long term measurements, especially for small turbine manufacturers. This thesis investigates four topics that could greatly impact the performance of a wind turbine: 1. the wind resource, and how by the spectral modelling of its unsteadiness it is possible to design better turbines and control systems, to adequately react in gusty wind conditions in order to maximise energy harvesting; 2. resource assessment using short-term measurements of wind, to reduce the uncertainty in annual energy production; 3. loss modelling in the generator and power converter to optimise overall efficiency of a wind energy conversion system; and 4. aerodynamic performance identification based on field measurements. On each of these fields, new methods are proposed and validated to improve the existing knowledge. Regarding loss modelling and optimisation, an algorithm to find the global optimum in a system with losses is proposed and tested in an 800 W vertical-axis wind turbine owned by Airborne- Energy Ltd. The experimental data collected in the field, confirms the validity of the approach and its ability to find the optimum despite the high inertia of the turbine, and the unsteadiness of the wind. It is shown that a control algorithm that seeks a global optimum, can increase the overall efficiency of the system, and reduces internal stress on the shaft and power electronics. Finally, as the optimisation algorithm developed in this thesis features the estimation of parameters for a turbine, the processed information can have two positive impacts for further study: detection of system faults and diagnostic of the health-state of the system; and, design specifications, as the manufacturer can have direct inputs of the performance of the turbine to make further improvements for new designs.