Prospective memory tasks which were assigned to the subjects during the course of experiments can be divided into two kinds according to its property: natural task and artificial task. And both task can be presented on a laboratory environment or an out laboratory environment. So there are four paradigms of tasks in prospective memory experiments: artificial task in natural situation (type 1), natural task in natural situation (type 2), artificial task in laboratory situation (type 3) and natural task in laboratory situation (type 4). Tests were done in turns to 47 undergraduates between bringing paper and pen (type 2), prospective memory test in laboratory (type 3), signing the date when finishing the test (type 4) and sending a text message to the experimenter 2 days later (type 1). The results indicated that prospective memories of different attribute in the same situation were consistent, especially in the natural situation. But the same attribute of prospective memories in different task situations were inconsistent. So prospective memories of the same task attribute in different task situations are inconsistent, prospective memories in laboratory and in natural situations can’t be anticipated with each other. The results could not only afford reference on improving ecological validity and choosing an appropriate experiment task of prospective memory research, but also provide illumination to the measurement of the ability of prospective memory.