We propose a method for resolving spectroscopic reaction monitoring data into concentration profiles and spectra for the pure chemical components. It can be used if a valid mechanistic model for the reaction can be postulated. In the analysis, the parameters of the model, i.e. the rate constants, are also determined. If several mechanistic models are possible, these can be tested and the model best describing the data can be selected. The method is based on refinement of concentration profiles from the mechanistic model by target testing. Subsequent rotation of the loadings from principal component analysis (PCA) yields the pure component spectra. The only input used is the spectroscopic data, the mechanistic model(s) to be tested/fitted and possibly, but not necessarily, the initial concentrations of the species participating in the reaction. The performance of the method has been evaluated for synthetic and experimental data described by some common and simple reaction models. A comparison with techniques for non-linear least squares fitting and self-modelling curve resolution is included in the text. (C) 1998 Elsevier Science B.V. All rights reserved.