Justice research has evolved by elucidating the factors that affect justice evaluations, as well as their consequences. Unfortunately, few researchers have paid attention to the pattern of rewards over time as a predictor of justice evaluations. There are two main objectives of this research. First, it aims to test the effect of reward stability on justice evaluations. Based on justice theory and prospect theory, we assume that an under-reward at one time cannot be fully offset by an equivalent over-reward at another time. Therefore, in unstable reward systems the asymmetry of the effect of unjust rewards with opposite directions will produce a lower level of justice evaluations over time. The second objective of this research is to show the moderating effect of the presentation order (primacy vs. recency) of unstable rewards on justice evaluations. The results from a controlled experiment with five conditions, which presents the instability of rewards in different orders, confirm both the negative effect of unstable rewards and the stronger effect of primacy on justice evaluations.