Spectroscopic parallaxes are an important tool for determining distances of individual stars in the Galaxy. Unfortunately, for many astrophysical applications this method does not produce sufficiently accurate results for distant stars. Therefore, we started an attempt to improve the accuracy of spectroscopic parallaxes by evaluating spectroscopic temperature and luminosity criteria such as those of the MK classification spectrograms which were analyzed automatically by means of a suitable computer program. If necessary, the grain noise is reduced by Fourier filtering techniques. Then, the computer derives an approximate spectral type and the wave length scale from selected prominent spectral features. As a next step the computer searches the spectrum for adsorption features and produces an internal catalog of detectable absorption lines. Then, depending on the approximate spectral type, line ratios, line widths, and line depths of selected lines are calculated and compared to the corresponding values of standard stars. Since the investigation is still in progress, no final result on the achievable accuracy can be given here. However, preliminary results indicate that spectroscopic parallaxes derived by the computer method are at least as accurate as those derived from the conventional visual classification technique. A more detailed account of this investigation will be published elsewhere.