The occurrence of a factor in soybean which deters feeding of the silkworm was demonstrated. The deterrent was not extractable with ether or ethanol, but was extracted with 90% ethanol or hot water. Preliminary experiments showed that the deterrent was adsorbed on charcoal, but neither on activated alumina nor in a steam distillate. Autoclaving did not inactivate the deleterious effect. A crude extract of the deterrent showed a marked negative feeding response by including 0·4 mg/g of the dry diet. Inclusion of the crude extract in the diet retarded the growth of larvae, whereas larvae grew well when receiving the 90% ethanol-washed meal. Feeding response for various soybean meals and proteins were compared, and better larval growth has generally obtained when the feeding response was more positive. It is, however, inconclusive whether or not the deleterious principle possesses some physiologically toxic effect. The feeding behaviour of newly hatched larvae was observed in the presence of favourable and/or unfavourable diets, and a brief discussion is made concerning the feeding behaviour of this insect.