We first present a brief discussion of early and current models (feeding systems) used to estimate animal energy and protein requirements and to predict performance based on feed composition and intake. We then touch on some limitations inherent in these systems. Next, we propose that dynamic (i.e. time-variant) models and mechanistic models (i.e. equations based on knowledge of physiological functions) are superior to earlier systems for both practical and research applications. Finally, we describe a number of applied and research models of animal growth and lactation to illustrate how biological concepts may be represented in equation form. Equations that represent current knowledge of underlying functions can be used to improve predictions of animal requirements and performance and to evaluate hypotheses about nutritional and physiological mechanisms that influence animal performance.