Retrospective self-reports are widely used to measure affect and well-being. But researchers have long assumed that people overestimate affective experiences in retrospect and that retrospective self-reports are thus biased. This is because of the memory-experience gap, a phenomenon in which retrospective ratings for a longer timeframe are higher than the average of repeated ratings for shorter timeframes. This discrepancy is the basis of theories about how some people may overestimate their past feelings in general and a bias in retrospective self-reports. Rather than reflecting a memory bias, however, the discrepancy could be due to differences in how people summarize their feelings over different timeframes. To remove this confound, we used an online convenience sample and measured affect over several timeframes for a week (N = 399; collected in 2022), as well as memory for past affect over the same timeframe. Longer timeframes (e.g., 1 week) were rated higher than shorter timeframes (e.g., averaged across each day of that week) for both negative and positive affect, demonstrating the memory-experience gap. But ratings for each day given at the end of the week, from memory, were lower than those given for each day during the week. Ratings based solely on memory were therefore in the opposite direction to the memory-experience gap. This brings into question the assumption made by some researchers that the “memory-experience gap” reflects a memory bias in retrospective self-reports. Generalizability to other methodological designs, constructs, and populations requires testing.