Weather radars provide valuable information on precipitation in the atmosphere but due to the way radars work, not only precipitation is observed by the weather radar. Weather radar clutter, echoes from non-precipitating targets, occur frequently in the data, resulting in lowered data quality. Especially in the application of weather radar data in quantitative precipitation estimation and forecasting a high data quality is important. Clutter detection is one of the key components in achieving this goal. This thesis presents three methods for detection of clutter. The methods use supervised classification and use a range of different techniques and input data. The first method uses external information from multispectral satellite images to detect clutter. The information in the visual, near-infrared, and infrared parts of the spectrum can be used to distinguish between cloud and cloud-free areas and precipitating and non-precipitating clouds. Another method uses the difference in the motion field of clutter and precipitation measured between two radar images. Furthermore, the direction of the wind field extracted from a weather model is used. The third method uses information about the refractive index of the atmosphere as extracted from a numerical weather prediction model to predict the propagation path of the radar’s electromagnetic energy. This facilitates the prediction of areas of clutter caused by anomalous propagation of the radar’s rays. The methods are evaluated using a large independent test set, and to illustrate the performance on individual radar images three typical case examples are also evaluated. The results of the evaluation of the methods show that each method has good skill in detection of clutter with an average classification accuracy of 95 %. The methods thus have the potential for increasing the quality of weather radar data in their operational use.