The inborn error of metabolism phenylketonuria (PKU, OMIM 261600) is most often due to inactivation of phenylalanine hydroxylase (PAH), which converts phenylalanine (Phe) into tyrosine (Tyr). The reduced PAH activity increases blood concentration of phenylalanine and urine levels of phenylpyruvate. Flux balance analysis (FBA) of a single‐compartment model of PKU predicts that maximum growth rate should be reduced unless Tyr is supplemented. However, the PKU phenotype is lack of development of brain function specifically, and Phe reduction rather than Tyr supplementation cures the disease. Phe and Tyr cross the blood–brain barrier (BBB) through the aromatic amino acid transporter implying that the two transport reactions interact. However, FBA does not accommodate such competitive interactions. We here report on an extension to FBA that enables it to deal with such interactions. We built a three‐compartment model, made the common transport across the BBB explicit, and included dopamine and serotonin synthesis as parts of the brain function to be delivered by FBA. With these ramifications, FBA of the genome‐scale metabolic model extended to three compartments does explain that (i) the disease is brain specific, (ii) phenylpyruvate in urine is a biomarker, (iii) excess of blood‐phenylalanine rather than shortage of blood‐tyrosine causes brain pathology, and (iv) Phe deprivation is the better therapy. The new approach also suggests (v) explanations for differences in pathology between individuals with the same PAH inactivation, and (vi) interference of disease and therapy with the functioning of other neurotransmitters.