Euler deconvolution is a widely used semiautomatic method for potential field data, but it produces sprays of solutions which complicate interpretation. It is demonstrated here that the accuracy of the solution locations can be improved if the method is applied iteratively. Euler deconvolution uses a window of datapoints to solve for the horizontal and vertical distances from the centre of the window to the potential field source. When applied iteratively the method uses the horizontal location of a solution to define the centre of a new window of data points. This new window of points is then used to generate an improved Euler solution, and the procedure is repeated. Solutions that are initially poorly located either migrate closer to potential field sources or diverge beyond the limits of the data, thus removing them from the result. The method is demonstrated on synthetic datasets and aeromagnetic data from South Africa.