The work focuses on the development of an artificial neural network (ANN) based model that can describe the adsorption of benzalkonium chloride from aqueous solutions onto commercially available kitchen paper. Various ANN architectures were tested in order to find the most suitable one in terms of overlapping between calculated and measured output data (coefficient of determination and mean absolute percentage error), as well as correctly interpolating outputs when using inputs form inside the experimental training range. The networks all had 4 inputs and 1 output, as well as a single hidden layer. Optimal ANN design was sought by varying both the number of neurons in the hidden layer and the type of transfer function towards it. The best find was employed in assessing the relative importance of input parameter values in the output, as well as the model’s suitability for predictions outside the training range.