Currently, obesity treatment rests on the "calories-in, calories-out" (CICO) rule, formally named the energy balance theory (EBT). It maintains that body weight (BW) increases as food calories are greater than expended calories but decreases when the opposite occurs; hence, weight stability is expected at energy balance meaning that over time energy-in equals energy-out. It follows that dietary regimens with identical energy content should evoke similar amounts of weight and fat loss with only minor differences that follow from diet's macronutrient composition, e.g., diet-induced glycogen depletion and water excretion. A vast collection of evidence shows, however, that low-carbohydrate diets typically result in much greater weight and fat loss than isocaloric low-fat diets. Furthermore, our recent mathematical analysis demonstrated that weight stability coincides with a persistent energy imbalance and not otherwise. As an alternative, the mass balance model (MBM) was proposed that fitted weight loss data and explained the often superior weight loss evoked by low-carbohydrate diets versus low-fat diets. Here, we expand on these observations by computationally contrasting the predictions of both models in two scenarios described in the literature: altering the diet's macronutrient composition while energy intake is kept at weight maintenance level; and the weight loss response as diet composition is changed under untreated type 1 diabetes. Our results indicate that MBM predictions are remarkably accurate while those of the EBT are clearly erroneous. These findings may represent the beginning of a paradigm shift in obesity research.