Chromatographic analysis of tobacco or tobacco smoke results in the identification of many compounds present in the samples. The use of high resolution chromatographic techniques such as GCxGC analysis can produce lists of tens of thousands of compounds present in a single sample, giving rise to the problem of extracting useful information from all the data. With mass spectral data, compounds can be identified according to compound classes. Some classes of compounds, such as fatty acid methyl esters, are easily identified by the presence of specific masses in the spectrum and the lack of significant abundance of other molecular fragments. Other compounds, such as chlorinated compounds are readily identified by the characteristic isotope cluster shown for the molecular ion, once the molecular ion is located in the spectrum. In other cases, such filters may lack specificity, but the structure of a GCxGC chromatogram tends to localize compounds by class, so the use of automatic spectral identification may be used with selection by location in the chromatographic plane to select the compounds of interest. This presentation shows the application of such automatic identification to locate compounds of classes in which identities might be desired or in which only summary information is needed. Chromatograms of smoke are automatically filtered for chlorinecontaining compounds and results for individual compounds are provided. Summary data is obtained from smoke samples, giving estimates of total concentration of compounds by classes such as fatty acid methyl esters, methyl ketones, benzene aromatics and other compounds, which can be used in the comparison of one type of cigarette against another.