The impregnation of salicylic acid following a supercritical fluid technology was determined experimentally for various adsorbing matrices such as Florisil, zeolite, silica gel and Amberlite under a wide range of experimental conditions. A thorough characterisation of these supercritical adsorption processes requires a large number of runs under various experimental conditions in order to cover all possible sources of variance, which may result in a tedious and time-consuming task. The overall information contained in the data set generated in this series of experimental runs can be analysed more efficiently using chemometrics. Here, multivariate data consisting of adsorption values determined for each material were analysed with principal component analysis and partial least-squares regression (PLS). The corresponding PLS models were subsequently applied to predict the impregnated amounts for other experimental conditions. Results showed a reasonable concordance between the experimental and predicted values.