Improved approaches to the problem of heterozygote detection for phenylketonuria (PKU) were developed in this study. The discrimination was based on 85 obligate heterozygotes and 45 controls who were neither pregnant nor on birth control medication. The best separation between hetrozygotes and normals was achieved with a linear discriminant function involving the logarithms of the serum concentrations of phenylalanine, tyrosine, and tryptophan. The theoretical overlap area between the distributions of heterozygotes and controls based on the above function, was 3.75%. In the 19 obligate hetrozygotes and 13 controls who were either pregnant or on birth control medication, the best separation was achieved with a linear discriminant function involving the logarithms of the serum concentrations of phenylalanine and tyrosine. The theoretical overlap area was 8.23%. The genetic accuracy of the discriminant function was confirmed by testing the results with parental-child exclusions, segregation analysis, and the frequency of heterozygosity in nonrelated collateral spouses. Finally, there was evidence suggesting that the antihypertensive agent, aldomet, alters serum tyrosine and tryptophan levels.