Records of 46 patients with classic phenylketonuria (PKU) were used to determine treatment variables associated with intellectual outcome. Patients comprised three groups: phenylalanine-restricted diet started 1) after 3 months and loss of diet control at a mean age of 7 years, 2) before age 3 months and loss of diet control at a mean age of 5 years, and 3) before age 3 months and through a mean age of 11 years. All underwent IQ testing during the diet; groups 1 and 2 were retested at a mean of six years off the diet. On the diet, groups 2 and 3 had higher IQs than group 1; group 3 IQ was also higher than IQ off diet in groups 1 and 2. After discontinuing the diet, group 2 IQs decreased significantly. Predictors of IQ in group 1 were age at loss of diet control and percentage of phenylalanine concentrations > 15 mg/dL; in group 2, mean phenylalanine concentrations and age at loss of diet control. Predictors of changes in group 1 IQs were global degree of dietary control and percentage of phenylalanine concentrations > 15 mg/dL; in group 2, phenylalanine concentrations of < 3 mg/dL and age at start of diet. Group 1 patients with phenylalanine concentrations < 3 mg/dL or > 15 mg/dL achieved no IQ gain by continuing the diet after age 7 years. Thus, intellectual prognosis is best for PKU patients who start a phenylalanine-restricted diet early and continue through age 12 years.