This paper will analyze data and concepts which underlie estimates of protein allowances for premature infants. It is fitting that this be done at Cornell University Medical School where careful metabolic and clinical studies have supplied the basis for currently popular, moderately "high" protein, moderately "low" fat feeding mixtures long advocated by Dutch and American pediatricians. It is important to distinguish between "allowances" and "requirements." "Allowances" is a permissive term implying nutrient levels that will cover individual variations both constitutional and environmental. "Requirements" is a tyrannical term implying absolute needs for all infants. Even its expression as a range or as a mean with a standard deviation does not jibe comfortably with practice of a pediatrician who is impressed daily with the variation in appearance, behavior, and adaptability of his patients. Socioeconomic or geographic factors may necessitate attempts to estimate requirements, but adaptability to minimal intakes under limited circumstances is