This paper explores the flight of small fixed-wing Unmanned Aerial Vehicle (UAV) in a non-steady environment. The vulnerability of light airplanes to wind is analyzed and the effect of such perturbations on airplane performance is incorporated in the equations of motion. A straightforward wind computation approach, which relies on the difference between the predicted motion of the aircraft and the real motion measured by sensors, is presented in order to be used for a path following application. The analysis takes into account the effect of the noise in sensors measurements and in estimates of orientation and airspeed components. One approach to reducing noise in wind estimates is proposed based on on-line adaptation techniques. Parameter estimation with minimum-order design is obtained using tuning functions. Simulations are carried out representing real flight scenarios in which the wind field is not constant and the sensor measurements are imperfect.