Delayed responses are a common phenomenon in experience sampling studies. Yet no consensus exists on whether they should be excluded from the analysis or what the threshold for exclusion should be. Delayed responses could introduce bias, but previous investigations of systematic differences between delayed and timely responses have offered unclear results. To investigate differences as a function of delay, we conducted secondary analyses of nine paper and pencil based experience sampling studies including 1,528 individuals with different clinical statuses. In all participants, there were significant decreases in positive and increases in negative affect as a function of delay. In addition, delayed answers of participants without depression showed higher within-person variability and an initial strengthening in the relationships between contextual stress and affect. Participants with depression mostly showed the opposite pattern. Delayed responses seem qualitatively different from timely responses. Further research is needed to understand the mechanisms underlying these differences.