An important factor that influences memory performance is the length of the delay between encoding new information into memory and the moment that information is to be retrieved. Recently, a small but growing body of research has examined the effect of the delay between intention formation and retrieval on prospective memory (PM) performance. These studies found that the effect of delay on PM is mixed. However, most of the researches are related to event-based PM. We knew little about the influence of delay on time-based PM. The present study explored the effect of retention intervals on time-based PM with two experiments. In experiment 1, the duration of ongoing task prior to the presentation of the first PM goal was manipulated. Participants were told to press the A key every 2 min or 8 min or 15 min when they were performing the ongoing task. The participants in control condition only performed the ongoing task about 7min as the 2 min delay condition. There were three PM responses in the PM conditions. In experiment 2, participants were required to perform filler task before ongoing task in which PM goals were embedded. The duration of filler task was 2 min or 15 min. The duration between the onset of ongoing task and the first PM goal was also 2 min or 15 min. The experiment conformed to a 2(filler task duration: 2 min vs. 15 min)×2(ongoing task interval: 2 min vs. 15 min) between-subjects design. For example, the subjects should perform 2 min filler task, then perform the ongoing task. They needed to press the A key every 15 min when they started the ongoing task. There were two PM responses. The results showed that the PM performance in 15 min delay condition was significantly lower than other two conditions when participants only carried out ongoing task in experiment 1. But, the reaction time (RT) in 2 min delay condition was longer than the other conditions. However, when the retention interval was defined by both the filler task duration and the ongoing task duration in experiment 2, the PM performance was different. The filler task duration by ongoing task interval interaction was significant. The subsequent analysis revealed that in the long filler task conditions, the PM performance in 2 min ongoing task interval was better than 15 min delay condition. In short filler task conditions, there was no significant difference between 2 min and 15 min ongoing task delay conditions. In addition, for the RTs of ongoing task, the ANOVA analysis revealed a main effect of ongoing task interval. The RTs for the short ongoing task interval were longer compared to the long ongoing task interval, especially when participants performed the short version of the filler task. The frequency of clock checking was positively associated with PM performance and participants increased their frequency of time monitoring closer to the target time in the two experiments. The present study suggests that the duration of filler tasks vs. ongoing tasks is an important factor to consider in evaluating the impact of delays between intention formation and intention retrieval. The delay of filler task and ongoing task may have different effects on intention maintaining in time-based PM. Filler task more or less alleviates the negative influence of long ongoing task delay on PM. Participants may use different strategies to maintain PM goal. However, many other factors will affect the intention retrieval in time-based PM, for example, the number and feature of filler tasks. The future research may consider these issues.