<p>The common assumption that depression is primarily related to reduced average well-being may obscure a more critical pattern. Drawing on cognitive reactivity theories of depression, we hypothesized that depression is mainly associated with the worst levels of well-being individuals experience, rather than with their average or peak well-being experiences. This preregistered hypothesis was tested by analyzing 98,146 momentary well-being reports provided by 1,468 participants from various regions of the world (i.e., Europe, Australia, United States of America), all of whom also completed a depression disposition questionnaire. Using linear mixed-effects modeling with bootstrapped confidence intervals based on 1,000 resamples, we found that the negative relationship between depression and well-being was stronger at the lowest quantiles of the well-being distribution. As hypothesized, the predicted well-being gap between individuals with low (− 1 <i>SD</i>) versus high (+ 1 <i>SD</i>) depression levels was significantly larger in the lower part of the well-being distribution than in its higher parts. Specifically, at the 10th quantile, the gap reached 0.71 standard deviations (99.9% CI [0.63, 0.80]), decreased to 0.58 at the median (99.9% CI [0.50, 0.67]), and was even smaller at the 90th quantile (<i>Δ</i> = 0.35, 99.9% CI [0.29, 0.40]). In other words, the strongest well-being differences between individuals with varying depression levels arise during their worst moments, whereas their typical or more favorable moments appear more similar. Theoretical and practical implications of these findings are discussed.</p>