Quantitative multicomponent analysis in UV-VIS spectroscopy can allow to replace high-performance liquid chromatographic (HPLC) methods or support them by increasing detection certainty. However, commercial software offers the possibility to carry out problematic analyses, because the potential of the individual multicomponent procedures is seldom exploited to full advantage. Usually, only one method is offered, without any modification possibilities. Difficult analysis may be caused by high concentrations and interferences between the compounds of the mixture. Thus the electron systems of the molecules respectively the UV spectra are altered in comparison with the spectra of the pure components in high dilution. These non-linearities may be compensated by multicalibration techniques with respect to the interpretation of multicomponent spectra, a possibility also existing for ordinary least squares (the K matrix approach). Spectral similarities between the individual calibration spectra shift non-linearities into the range of characteristic differences between the components. It is therefore useful to make a selection of the examined wavelengths to exclude similar ranges from the calculation of the concentrations. When the more common least squares methods are used, impurities or unknown components lead to wrong parameters (concentrations), and the procedure does not provide for correcting the calibration model. On the assumption that the above-mentioned faults can becombined in a single component, the adaptive Kalman filtering makes a proposal for the unmodelled compound [l]. This proposition can be used to alter the model and run the procedure a second time. The non-adaptive form of the filter is