Although blood plasma has been used to diagnose diseases and to evaluate physiological conditions, it is not easy to establish a global normal concentration range for the targeting components in the plasma due to the inherent metabolic diversity. We show here that NMR spectroscopy coupled with principal component analysis (PCA) may provide a useful method for quantitatively characterizing the metabolic diversity of human blood plasma. We analyzed 70 human blood plasma samples with and without addition of ibuprofen. By defining the PC score values as diversity index (I(div)) and the drug-induced PC score value change as interaction index (I(dist)), we find that the two indexes are highly correlated (P < 0.0001). Triglycerides, choline-containing phospholipids, lactate, and pyruvate are associated with both indexes (P < 0.0001), respectively. In addition, a significant amount of lactate and pyruvate are in the NMR "invisible" bound forms and can be replaced by ibuprofen. The diffusion and transverse relaxation time weighted NMR approaches gave rise to a better characterization of the diversity and the interaction than that of the one acquired using NOESYPR1D with 100 ms mixing time. These results might be useful for understanding the blood plasma-drug interaction and personalized therapy.