Extract: Phenylalanine (100 mg/kg) was administered orally to 19 parents of patients with phenylketonuria and to 26 control subjects. Concentrations of phenylalanine and tyrosine in plasma were determined by column chromatography before phenylalanine administration and at 1, 2, 3, and 4 hours after loading. The values of two control subjects (C 21 and C 25) and of two parents (P 4 and P 6) were unusually low or high. Excluding subjects C21, C25, P4, and P6, the following calculations of the values after loading were made: Mean and SD of mean of phenylalanine levels at each time: in spite of a significant difference, there was a slight overlap. Sum of the phenylalanine values: significant difference, no overlap. Sum of the phenylalanine/tyrosine ratios: significant difference, no overlap. Using these criteria, control subjects C21 and C25 were considered to be heterozygotes. The values of subject P 4 (father of a known patient) were within the range of the control group; paternity was excluded by blood group analysis. The values of subject P 6 were unusually high and could not be explained satisfactorily. The concentrations of phenylalanine and tyrosine after loading were evaluated by discriminatory analysis. Thus, subjects C 21, C 25, and P 4 were excluded, but values were calculated both including and excluding subject P 6. To compare the different criteria of evaluation, the index of Penrose (discriminatory power = D/s) was calculated. A better separation was obtained with discriminatory analysis than with the sum of the phenylalanine values or of the phenylalanine/tyrosine ratios. The use of column chromatography and evaluation by discriminatory analysis seems to improve the possibilities of detection of heterozygotes for phenylketonuria. Speculation: With discriminatory analysis, the separation of heterozygotes for phenylketonuria from normal subjects was better than with other criteria of evaluation. We believe that this method could be used successfully for determination of heterozygosity in other inborn errors of metabolism with recessive inheritance when several variables are available for each subject examined.