An improved method of estimating molecular weights of volatile organic compounds from their mass spectra has been developed and implemented with an expert system. The method is based on the strong correlation of MAXMASS, the highest mass with an intensity of 5% of the base peak in a mass spectrum, with the true molecular weight. Linear corrections to MAXMASS, which yield molecular weight estimates, were derived empirically using the expert system. The system is based on a sequential modular design with a primary classifier linked to molecular weight estimators for the resulting six classes. These are non-halobenzenes; chlorobenzenes; bromoalkenes/alkenes; mono- and dichloroalkanes/alkenes; tri-, tetra- and pentachloro-alkanes/alkenes; and others. Performance tests were made with the old and improved expert systems and the Self Training Interpretive and Retrieval System (STIRS) on NIST reference spectra of the 107 training compounds; on a set of 32 test NIST reference spectra of other compounds found in air samples; on a set of 100 randomly selected NIST reference spectra; and on 27 spectra taken from actual field gas chromatography/mass spectrometry samples.