The Dual Mechanisms of Control (DMC) theory suggests that cognitive control consists of two modes: proactive control (anticipatory and sustained control during the cue encoding and maintenance periods) and reactive control (control is engaged after rather than before the probe and in a just-in-time manner rather than consistently). There is a tradeoff between these two kinds of control to optimize task performance. Several studies have shown that the reward context is associated with a shift towards proactive control. Recently, it is suggested that awareness of reward cue is not necessary for the cue to work. Reward cue can improve working memory and cognitive control even when it occurs beyond subjects’ consciousness. However, it is still unclear how subliminal reward modulates the tradeoff between proactive and reactive control. The present study conducted an experiment to investigate this question by using the AX-CPT task. In this study, participants were engaged in a reward version of AX-CPT task, in which some reward-related cues and feedbacks were introduced. These reward-related cues and feedbacks were presented supraliminally (284 ms followed by a mask for 16 ms) or subliminally (16 ms followed by a mask for 284 ms) before and after each cue-probe sequence of AX-CPT task. There were two kinds of reward-related cues (“$$$” indicating a 50 scores potential reward, “###” indicating no potential reward) and two corresponding feedbacks (“+50” and “+--”). In AX-CPT task, letters (“A”, “B”, “X”, and “Y”) were individually and sequentially displayed on a computer screen. A target response was required only when an ‘‘X’’ probe is preceded by an ‘‘A’’ cue (AX trials). All other probe stimuli (AY, BX, and BY trials) required non-target responses. Participants were told to complete baseline condition, supraliminal reward condition and subliminal reward condition one by one. In baseline condition, the reward-related cues were presented subliminally and participants performed the AX-CPT task without any instruction on financial incentives. During the two subsequent conditions, participants would gain some money if they respond correctly and quickly. The only difference between supraliminal and subliminal reward conditions was the presentation of reward-related cues and feedbacks, the former was supraliminal while the latter subliminal. Two-way analysis of variance (ANOVA) was used to analyze the error rate and RT. The behavioral results showed that supraliminal reward cues improved the performance of AX trials while subliminal reward cues had no effects. Nevertheless, the interactions between trial types (AY vs. BX) and reward conditions (baseline vs. supraliminal reward, or baseline vs. subliminal reward), which was thought to index the tradeoff between proactive and reactive control, was significant. It means that just like supraliminal reward cues, subliminal reward cues modulated the tradeoff between proactive and reactive control and led to a preference toward proactive control. In contrast to some recent results, the present study does develop the DMC theory and the research on subliminal reward.