The EISCAT VHF incoherent scatter radar (ISR) is capable of measuring the incoherent scatter spectrum down to upper stratospheric altitudes, during times of excess ionisation, such as produced for example by solar proton events. However, interpretation of measured ISR spectra as physical parameters is not straightforward and needs additional information either from other measurements or from theoretical models. Published results on variations of atmospheric composition in the mesosphere, deduced from ISR data, are extremely limited in number, most of them being also speculative in nature [1]. We present a detailed study of sunset variations in the concentrations of chemically important minor constituents, above Tromsø, Norway, during the solar proton event of October 23, 1989. The calculations are based on the recent development of the time dependent Sodankylä Ion Chemistry model (SIC) [2], which now includes coupled neutral and ion chemistry, dissociation of neutrals, vertical diffusion, a proper solar radiation model instead of a fixed reference as before, improved negative ion chemistry at lower altitudes and effects by secondary electrons, to mention a few improvements in the model. Proton flux data measured onboard the GOES satellite is used as input in the model, in order to calculate the ionisation rates of the neutrals in the altitude range from 50 to 100 km. Temperature is taken from the MSIS-90 model. The model results of ion composition are used to calculate a theoretical ISR spectrum and its width, which is compared with the experimental results from the EISCAT VHF radar, which was running the CP6 ionospheric D–region experiment during the solar proton event. Time behavior of the calculated ISR spectrum follows reasonably the experimental one at all altitudes. Deviations between the two are seen in the absolute values of the ISR spectral width. Del Pozo et al. [3] studied the same event with an earlier, purely ion chemistry version of the SIC model. The model in their study was not able to explain the time variations seen in ISR spectra. They deduced qualitatively that variations in the concentration of atomic oxygen must be responsible for the observed variations of ISR spectral width. In this paper, we present for the first time quantitative results for the time variation of concentrations of minor neutral species, including atomic oxygen.