In this paper a comparison of two column characterisation systems is reported: the method based on the hydrophobic-subtraction model of Dolan and Snyder (HS method) versus the method developed at the Katholieke Universiteit Leuven (KUL method). Comparison was done for seven different pharmaceutical separations (fluoxetine, gemcitabine, erythromycin, tetracycline, tetracaine, amlodipine and bisacodyl), using a set of 59 columns. A ranking was built based on an F value (KUL) or F(s) value (HS) versus a (virtual) reference column. Both methods showed similar probabilities of ranking patterns. Correlation of the respective test parameters of both approaches was poor. Both methods are not perfect and do not match well, but anyhow yield results which allow, with a relatively high certainty, the selection of similar or dissimilar columns as compared to a reference column. An analyst that uses either of the two methods will end up with a similar probability to choose an adequate column. From a practical point of view, it must be noted that the KUL method is easier to use.