[en] This paper describes work carried out within the IAEA EMRAS program NORM working group to test the predictions of three computer models against measured radionuclide concentrations resulting from discharges from a lignite power plant complex. This complex consists of two power plants with a total of five discharge stacks, situated approximately 2-5 kilometres from a city of approximately 10,000 inhabitants. Monthly measurements of mean wind speed and direction, dust loading, and ²³⁸U activities in fallout samples, as well as mean annual values of ²³²Th activity in the nearest city sampling sites were available for the study. The models used in the study were Pc-CREAM (a detailed impact assessment model), and COMPLY and CROM (screening models). In applying the models to this scenario it was noted that the meteorological data provided was not ideal for testing, and that a number of assumptions had to be made, particularly for the simpler models. However, taking the gaps and uncertainties in the data into account, the model predictions from PC-CREAM were generally in good agreement with the measured data, and the results from different models were also generally consistent with each other. However, the COMPLY predictions were generally lower than those from PC-CREAM. This is of concern, as the aim of a screening model (COMPLY) is to provide conservative estimates of contaminant concentrations. Further investigation of this problem is required. The general implications of the results for further model development are discussed. (author)