<jats:title>Summary</jats:title><jats:p>To analyze abnormal fertilization in a hereditary mosaic strain (<jats:italic>mo/mo</jats:italic>) of <jats:italic>Bombyx mori</jats:italic>, the percentages of diploidy mosaic, polyploidy mosaic and polyploid eggs in a batch were estimated by using egg colour mutants (<jats:italic>pe re</jats:italic>). Among 48 890 eggs from crosses of <jats:italic>pe</jats:italic> + / + <jats:italic>re, mo/mo</jats:italic> females with <jats:italic>pe re/pe re</jats:italic> males, 9409 abnormal eggs were obtained; 4472 of them were diploidy mosaics (red-white eggs), 4038 were polyploids (black eggs) and 899 were polyploidy mosaics (566 black-white, 256 black-red and 77 black-white-red eggs). The total number of diploidy mosaic eggs was approximately equal to that of polyploid eggs. A significant correlation was detected between the diploidy mosaic and polyploid egg ratios within a batch. This suggests that diploidy mosaics are produced by double fertilization in which two genetically non-identical egg nuclei are fertilized in turn by a sperm, and polyploids are formed by the fertilization of a diploid, non-disjunctive egg nucleus gamete by a single sperm. Our results also indicated the presence of common factors modifying both mosaic and polyploid frequency. The concordance of the observed ratio of polyploidy mosaic eggs (1·84%) with the expected value (diploidy mosaic ratio × polyploidy ratio × 2 = 1·83%) suggests that the formation of mosaics occurs independently of the formation of polyploids in this abnormal fertilization process. We point out that it is necessary to modify Goldschmidt & Katsuki's general model to explain abnormal fertilization, and we propose several possible models.</jats:p>