The results seem to suggest that the k-nearest neighbour classifier, operating on the first 26 principal components of the daily surface level pressure field, is a sensible approach to reproducing the Grosswetterlagen classification scheme for atmospheric circulation over the regional, European scale. The results obtained are slightly inferior to those previously obtained using a generalised linear model with a softmax output function, however the difference in overall accuracy is small. It should be noted that the generalised linear model is also more efficient in operation than a k-nearest neighbour classifier.