Although the 15 children in this study fit into the classic picture of anorexia nervosa, the title of the report is a misnomer, since none of the patients experienced marked anorexia. The failure to eat resulted from fear of eating, rather than loss of appetite. Previous workers are in agreement that this disorder represents a syndrome rather than a disease entity and that it may occur in people with various types of personalities and various degrees of ego strength. While many bizarre fantasies are associated in the patients' minds with the eating disturbance, the literature does not contain information about why, in these patients, such fantasies become associated with enough anxiety and guilt to require the defense of not eating. Our patients denied that they were starving themselves to the point of emaciation, were greatly preoccupied with food during their illnesses, and often served food to the other children on the ward. Preconscious and conscious fantasies relating to food and eating included animistic ideas about food, delusions that certain kinds of food were poisonous, fear of oral impregnation and gastric pregnancy, ideas of anal birth, orally aggressive and sometimes cannibalistic impulses, and the equation of not eating with a lifelong childlike dependent status. We had expected to find more evidence of repression of the fantasies connected with eating. We now believe isolation of affect was the chief defense mechanism that permitted such bizarre ideas to remain accessible to awareness. Open conflict between the mothers and grandmothers of many of our patients was a prominent factor during the patients' infancies. The children often recreated this conflict, using food as the bone of contention, while they were ill. The anxiety-laden fantasies about food and eating, preoccupation with food, fear of adult sexuality, and reluctance to give up the child's dependent status that were present in our patients often paralleled identical fantasies and attitudes of their parents. Contrary to our expectations, we obtained a history of early feeding disturbances in only 8 patients and have discussed the possible significance of this finding. The events that precipitated anorexia nervosa in these children do not differ from those associated with other psychological disorders. The connection between the precipitating events and the specific symptom of not eating only becomes evident in the context of the child's fantasies or the family myths shared by the child. None of these girls menstruated during anorexia; starvation is not the major factor in producing the amenorrhea. Our patients showed a high degree of anxiety about the menses and other sexual functions. The link between menstrual function and anorexia appeared most typically as amenorrhea pregnancy, or amenorrhea and not eating equals regression to babyhood. Manifest depression, present in all but 2 patients, differs from classic clinical depression. This finding is consistent with previous observations that children, in general, do not show classic depression. Efforts to formulate the anorexia syndrome exclusively in terms of the dynamics of depression have failed. Regression is as characteristic of anorexia nervosa as is depression. While regression is characteristic of children with psychiatric disorders in general, and also of children with physical illnesses, our patients seemed more deeply regressed than children with any other psychological disorder (except psychosis) and also more deeply regressed than other physically ill children. We have discussed how the particularly severe ambivalence present in the parents of these children may foster a particularly deep regression in the child. Lastly, we have noted some factors that seem related to the outcome of this syndrome.