A serious shortcoming of bibliometric studies based on the (Social) Science(s) Citation Index is the lack of an universally applicable subject classification scheme as individual papers are concerned. Subject classification of papers on the basis of assigning journals to subject categories (like those found in the various supplements of ISI databases) works well in case of highly specialised journals, but fails for multidisciplinary journals such as Nature,Science and PNAS and so far as subfields are taken into consideration - also for "general" journals (e.g. JACS or Angewandte Chemie). This study presents the results of a pilot project attempting to overcome this shortcoming by delimiting the subject of papers published in multidisciplinary and general journals by an item-by-item subject classification scheme, where assignment is based on the analysis of the subject classification of reference literature. The results clearly confirmed the conclusions of earlier studies by the authors in the field of reference analysis. For the really important journals (sufficiently high number of annual publications and high impact with respect to the field), the share of classifiable papers was surprisingly high, and the assignment proved reliable as well. Since papers in the leading general and multidisciplinary journals are frequently citing general and multidisciplinary journals, an iterated application of the procedure is expected to increase the number of classifiable publications.