In this article, we report on an experimental investigation of the spontaneous, uncued recall of prospective memories (memories for activities to be performed at a later time). Einstein & McDaniel (1990) developed an experimental paradigm for investigating prospective memory, in which participants were required to perform both an ongoing task (e.g., remembering words presented on the PC screen) and a prospective memory task (e.g., pressing a designated key whenever they saw a particular word, such as rake). They also claimed that there are two types of prospective memory: one is event-based prospective memory recall, which is triggered by another event ("I will give a message to John when I meet him"), and the other is time-based prospective memory recall, which is to be done after a particular period of time has elapsed (I will call Mary in 30 minutes") or at a certain time ("I will watch TV at 7:00 PM"). We examine the nature of the Einstein and McDaniel's paradigm and show that several important aspects of prospective memory have been left unstudied; specifically, spontaneous, uncued recall. We consider that it is caused by the cue-oriented nature of the paradigm. Furthermore, considering the prospective memory function in our everyday life, we cast doubt of the validity of the dissociation between time-based and event-based prospective memory. To investigate these two issues, we conducted a task-content oriented experiment which was a refined version of Einstein and McDaniel's paradigm. Thirteen undergraduates (9 male and 4 female) were presented 4 photographs on the PC screen simultaneously, and were required to judge which one of these four belonged to a different category (ongoing task). They were also required to stop the ongoing task when a photograph of envelopes was presented during the ongoing task and to call the experimenter in order to answer a questionnaire in an envelope before the experiment finished. Six participants were randomly assigned to an uncued condition, in which the photograph of envelopes was not actually presented (a photograph of a compass was presented instead) and 7 participants to a cued condition. The result showed that, although the expected recall cue was not presented, all the participants in uncued condition spontaneously remembered the prospective memory task. Furthermore, it was revealed that spontaneous recall did not occur randomly; instead, it frequently occurred near the end of the ongoing task, which is similar to the U-shaped clock-checking curve in the time-based prospective memory research (Ceci et al., 1988). These results suggest that participants in the uncued condition performed their event-based prospective memory task as a time-based one, and support our claim that prospective memory has both a time-based and event-based nature.