Cancer patients have the highest prevalence of malnutrition of any group of hospitalized patients. The potential causes of this malnutrition are numerous, as elements of both starvation and stress are evident in the cancer-bearing host. The presence of the tumor alone may lead to reduced intake of nutrients and treatment modalities of surgery, chemotherapy, and radiation therapy further exacerbate nutritional deficits. It is clear that the tumor requires energy substrates to grow, and that these substrates are exacted from the host. Animal studies identify progressive nutritional depletion concomitant with increasing tumor growth during ingestion of a regular diet. This appears predominantly due to reduced dietary intake in addition to host metabolic alterations. In animal/tumor models deliberate dietary protein depletion results in severe host weight loss, but also causes diminished tumor growth rates. Dietary manipulation in these animal/tumor models have demonstrated methods of improving tumor response to chemotherapy by manipulation of tumor growth rates. In addition, drug-pharmacokinetics have been altered by dietary manipulation. However, data from animal/tumor models are not directly applicable to man since the tumor in animals usually results in the death of the host within six to eight weeks. Nevertheless, controlled laboratory studies in animals provide basic metabolic information which promotes understanding of host/tumor relationships in man. In cancer patients malnutrition has prognostic value, leads to a distortion of body composition with erosion of body protein and fat stores, and compromises the delivery of adequate therapy. There is no direct objective evidence of accelerated tumor growth in humans with cancer who receive nutritional support as part of their treatment regimen. The host benefits to the extent that body composition is at least maintained during the period of nutritional repletion. Thus, nutritional support provides support to the patient during periods of treatment and dietary deprivation. No improvement in the tumor's response to therapy, however, has been demonstrated by this approach.