In this paper, we show the results from four different experiments in which sea bass, maintained under laboratory conditions, could choose between two or three different diets through self-feeders, which gradually increased the complexity and potential range of selection, to design their own diet in accordance with their requirements. At first, sea bass were allowed to select between two complete diets differing in the proportion of protein (52-58%): this showed their capacity to distinguish between two diets made of the same ingredients. Next, two incomplete diets, containing a fixed amount of protein (56%) and lacking either fat or carbohydrate, were made available. Three mixed diets made up of pairs of macronutrients (protein-carbohydrate, protein-fat or fat-carbohydrate) were tested in the next experiment and, finally, three diets containing only one macronutrient (protein fat or carbohydrate) were made available to fish. Taking into account selection made by the fish in the first three experiments, in which macronutrient selection was statistically different, protein was the main macronutrient chosen by fish (278.15 kJ/kgBW/day, on average), followed by fat and carbohydrate (162.85 and 64.56 kJ/kgBW/day, respectively). In conclusion, the results reveal the ability of sea bass to select an appropriate diet from experimental diets containing two or three macronutrients and suggest that the proposed methodology is a powerful tool for studying the differing nutritional needs of different species of fish.