In this study, the neurovascular coupling relationship was noninvasively studied in the human visual cortex. Graded neuronal/hemodynamic suppression conditions were generated using a paired-stimulus paradigm. Visual evoked potential was measured to quantify neuronal activity. Hemodynamic activities were measured and quantified by perfusion and blood oxygenation level-dependent changes. All quantification was normalized to the same activation condition induced by a single stimulus paradigm within each experimental session. This experiment design eliminated the confounding factors such as anesthesia and inconsistent neurovascular coupling patterns within and/or among tasks. The results reveal that (i) there is a tight neurovascular coupling at graded neuronal suppression conditions; (ii) the neurovascular coupling relationship contains a subtle, but significant, nonlinear component; and (iii) the linear model, nevertheless, is still a good approximation reflecting the neurovascular coupling relationship. This study extends the range of the neurovascular coupling relationship from graded neuronal excitation conditions to graded neuronal suppression conditions.