That distinct enzyme activities in an unbranched metabolic pathway are evolutionarily tuned to a single functional requirement is a pervasive assumption. Here we test this assumption by examining the activities of two consecutively acting enzymes in human erythrocytes with an approach to quantitative evolutionary design that avoids the above-mentioned assumption. We previously found that avoidance of NADPH depletion during the pulses of oxidative load to which erythrocytes are normally exposed is the main functional requirement mediating selection for high glucose-6-phosphate dehydrogenase activity. In the present study, we find that, in contrast, the maintenance of oxidized glutathione at low concentrations is the main functional requirement mediating selection for high glutathione reductase activity. The results in this case show that, contrary to the assumption of a single functional requirement, natural selection for the normal activities of the distinct enzymes in the pathway is mediated by different requirements. On the other hand, the results agree with the more general principles that underlie our approach. Namely, that (i) the values of biochemical parameters evolve so as to fulfill the various performance requirements that are relevant to achieve high fitness, and (ii) these performance requirements can be inferred from quantitative systems theory considerations, informed by knowledge of specific aspects of the biochemistry, physiology, genetics, and ecology of the organism.human erythrocytes ͉ oxidative stress ͉ safety factors ͉ quantitative evolutionary design ͉ systems biology A lthough numerous mutations can drastically change the values of biochemical parameters such as enzyme activities, these values are often narrowly distributed in natural populations. This outcome owes largely to the interplay between constraints imposed by physicochemical laws and natural selection for good performance of the organism's biochemical circuits. The strength of natural selection in shaping the design of living organisms at the molecular level (1) is just beginning to be appreciated. In humans, natural selection is strong enough to almost prevent fixation of mutations that decrease fitness by Ͼ0.002%. Furthermore, Ͼ70% of the amino acid-changing mutations are selected against (2, 3). In the many organisms that have larger effective populations, even smaller fitness differentials can drive natural selection. Understanding the quantitative design for biochemical parameters that has been achieved through natural selection (i.e., the qualitative and quantitative functional requirements that mediate their evolutionary tuning) is a goal of systems biology with far-reaching implications. Such understanding would provide valuable insight regarding physiological as well as evolutionary adaptation (4, 5) and guidance for metabolic engineering (6). However, this type of understanding for parameters such as enzyme activities remains limited.The assumption that the distinct enzyme activities in a metabolic pathway are evolutio...