In this article, we report on an experiment designed to improve the learning of metabolic biochemistry by nutrition and medical undergraduate students. Twelve students participated in a monitored lunch and had their blood extracted for analysis 1) before lunch, 2) 30 min after lunch, and 3) 3 h after lunch. The subjects were divided in two groups. One group had a hyperglicidic meal [pasta plus orange juice: 80% carbohydrate, 10% protein, and 10% lipid (estimated values)] and the other group had a hyperlipidic meal (calabresi pizza plus diet soda: 36% carbohydrate, 18% protein, and 46% lipid). Individual quantities of food were based on body mass index, age, and sex. The blood parameters analyzed were glucose, triglycerides (TG), and urea. Glucose remained constant in the three measurements in both groups. The TG concentration in the pasta group was constant before and after lunch but increased significantly during the evening. In the pizza group, TG increased after lunch and remained constant in the evening. Levels of urea increased only in the evening, specially in the pizza group. These results were used for the final biochemistry exam. With the maximum score set as 10, the average score was 6.0 +/- 2.4 (n = 102). We considered this activity a unique way of evaluating important issues on metabolism, because students had several hours to work on the final exam (with free access to a bibliography). It was also a good didactic experience (problem-based learning like) for the subject students, because they had to work in all phases of the experiment (idealization, realization, and analysis) and participated actively in the elaboration and correction of the exam.