The main results were: - Essential contributions have been made to an improved noise emission measurement method published as IEA recommendations . In particular, methods have been investigated for reducing wind induced microphone noise. A special secondary wind screen has been designed and recommended for practical measurements and it has been shown that a vertical board can be used to the improve signal-to-noise ratio. - A new noise propagation model has been developed, validated and programmed, based on data from measurement and calculation of wind turbine noise propagation carried out within the present and an earlier project. Noise attenuation in 1/3-octave bands is calculated. For most practical calculations the ground effect on the overall A-weighted noise level of a wind turbine can be assumed to be between +1 and +2 dB, as an approximation. Calculated and measured attenuation of total A-weighted noise levels are estimated to agree within 1 or 2 dB in these cases, while larger inaccuracy should be expected in other cases, especially in shadow zones. - Round Robin testing showed that sound power levels determined simultaneously by different partners were within approximately 0.5 dB, while measured tonality differed by up to 10 dB, primarily due to variation in partners' interpretation of the method. The latter implies that the tonality measurement method should be further developed and tested once procedures are defined in more detail. - The reproducibility of the measured overall A-weighted sound power level for a given make and model of wind turbine is in the order of 1.5 dB, expressed as one standard deviation, when the wind turbine is situated in flat terrain. In undulating terrain the measured sound power level may depend on how the wind speed is measured. Measuring wind speed 10 m above the ground can lead to unacceptable uncertainty in noise emission of tall wind turbines, and declaration of wind turbine noise should be based on wind speed calculated by means of the power curve. - Varying size and shape of ground boards led to less than 0.2 dB differences in measured sound power level, while variation in incident air turbulence and varying measurement distance had no measurable effect. Varying the wind speed sensor positions implied 0.5 dB deviations in sound power level of wind turbines in flat terrain. - A laboratory technique has been developed for assessing subjective unpleasantness of wind turbine noise. Noise tonality and noise fluctuation strength seemed to be the psycho-acoustic signal parameters best correlated with unpleasantness.