A flow injection analysis (FIA) system furnished with a gel-filtration chromatographic column and with photodiode-array detection was used for the generation of second-order data. The system presented is a model system in which the analytes are blue dextran, potassium hexacyanoferrate(III) and heparin. It is shown that the rank of the involved sample data matrices corresponds to the number of chemical components present in the sample. The PARAFAC (parallel factor analysis) algorithm combined with multiple linear regression and tri-PLS (tri-linear partial least-squares regression), which allows unknown substances to be present in the sample, are implemented for FIA systems and it is illustrated how these three-way algorithms can handle spectral interferents. The prediction ability of the two methods for pure two-component samples and also the predictions ability in the presence of unknown interferents are satisfactory. However, the predictions obtained by tri-PLS are slightly better than those obtained using PARAFAC regression algorithm.