Abstract Neither immunologic nor genetic concepts of carcinogenesis have yet been decisively confirmed, and epigenetic theories, as formulated so far, are either non-predictive or insufficiently consistent with morphologic and experimental evidence. Computing data, concerned with carcinogenic mechanisms and neoplastic changes at the level of the endoplasmic reticulum, may lead to a new coherent understanding of tumor pathogenesis. Carcinogenic agents initiate biophysical perturbations, chemical alterations and conformational transitions in the membrane lattice of the endoplasmic reticulum. Foremost among the resulting neoplastic changes is an increased, irreversible separation of polyribosomes from membranes of the ergastoplasm. The carcinogenic process, apparently, deletes a protein required for polysome attachment. Since microsomal cytochromes can be synthesized by membrane-bound polysomes only, the translation of genetic information for their biosynthesis is irreversibly restricted. A similar, self-perpetuating deficiency may be postulated for the polysome attachment protein. Activities, depending on cytochromes P-450 and b 5, are hampered, e.g. those of the monoxygenase system. Cholesterogenesis is derepressed. Ratios of phospholipids/cholesterol are decreased, and lipid-protein complexes, altered both in structure and function. Another distinct effect of the membrane-polysome separation is the unmasking of thiol-disulfide exchange enzymes which, in turn, stimulate the biosynthesis of proteins and of deoxyribonucleotides involved in cell replication.