We investigate a novel bias in online rating systems, “memory bias,” arising from the temporal gap between consumption and review publication. Through an analysis of 753,806 hotel reviews and a randomized experiment, we find: (1) Memory bias has a greater spillover effect on perceived performance than on expectations, boosting ratings. (2) The marginal influence of memory bias on online reviews demonstrates a U-shaped relationship with time. (3) Over time, the variability in memory bias gradually diminishes. (4) Memory bias in online reviews arises from individuals accessing different types of knowledge over time, mediated by the level of concreteness in the accessed information. These insights deepen our understanding of memory bias and provide valuable implications for stakeholders in tourism and hospitality.