In real-world automotive control, there are a lot of constraints to be considered. In order to explicitly treat the constraints, we introduce a model-prediction-based algorithm called a reference governor (RG). The RG generates modified references so that predicted future variables in a closed-loop system satisfy their constraints. One merit of introducing the RG is that effort required in control development and calibration would be reduced. In the preceding research work by Nakada et al., only a single reference case was considered. However, it is difficult to extendedly apply it to more complicated systems with multiple references such as the air path control of a diesel engine, which suffers from interference between boosting and exhaust gas recirculation (EGR) systems. Moreover, in the control, multiple constraints need to be considered to ensure hardware limits and control performance. Hence, it is quite beneficial to cultivate RG methodologies to deal with multiple references and constraints. In this paper, we develop the RG algorithm to allow for multiple references. The RG algorithm is composed of the following three steps: (i) predicted future variables are computed using a prediction model identified, (ii) an objective function describing the degree of constraint satisfaction over a prediction horizon is evaluated, and (iii) its minimization based on the two-dimensional gradient descent method results in the optimal modified references. We demonstrate the effectiveness of the present RG algorithm in a transient driving cycle experiment using a real engine, in which constraints are enforced on maximal boost pressure, turbine speed, compressor surge and maximal and minimal EGR rates. This experiment implies that we have expanded the applicability of an RG to system with multiple references compared to the previous work for only a single reference.