Background: Length of stay (LOS) has been used as a surrogate marker for patients’ well-being during hospital treatment. We systematically reviewed all pertinent literature on the role of nutritional status in predicting LOS in cancer. Methods: A systematic search of human studies published in English was conducted using the MEDLINE data base (all articles published as of December 2010). We searched using the terms ‘nutritional status’ and ‘nutritional assessment’ and ‘nutritional screening’ and ‘malnutrition’ in combination with the following terms: length of stay, length of hospital stay, duration of stay, and duration of hospitalization together with ‘cancer’ or ‘oncology’. Results: The MEDLINE search identified a total of 149 articles, of which only 21 met the selection criteria. Of the 21 studies, 10 studies investigated gastrointestinal cancer patients, 4 gynecological cancer, and 7 heterogeneous cancer. Eight studies used subjective global assessment (SGA) or patient-generated SGA (PG-SGA), 9 used serum albumin and/or BMI, and 4 used other methods of nutritional assessment. Conclusions: Validated nutritional tools such as SGA/PG-SGA are better predictors of LOS in gastrointestinal cancers requiring surgery than in nonsurgical gastrointestinal cancer patients. Correcting malnutrition may decrease the LOS and perhaps even lower the rate of hospital readmissions in this population.