The aim of this article is to compare the pretend theories of Lillard, A., Harris, P., & Perner, J.and to get the suggestions for the further researches of pretend play.To consider the issue of early understanding of pretend, Perner(1991)distnguished the acting as if paradigm from the symbolic function paradigm, in terms of the formar rather than the latter, against Leslie's(1987)Decoupling Theory, Perner et al(1994)proposed the Prelief hypothesis that under 4-year-old children cannot discriminate pretend from false belief but have a theory of Preliefinstead.To test this, Perner et al(1994)assessed children's ability to discriminate between pretence and belief and verified their hypothesis.For the analysis of Japanese children's speech acts(dialect and standard)in make-believe play, Kayo et al(1996)made a difference between the paradigm of make believe play(pretend play)-reality and imagination-reality distinction, and claimed that for the study of children's play activity the former distinction should be adopted.Hence, we agree with Perner(1991)'s acting as if paradigm rather than the symbolic function paradigm.However, Perner et al(1994)studied the "concept of pretence"rather than the "pretend action".In line with the symbolic function paradigm, Harris, P.and his colleagues studied mainly children's pretend action and found that 2-year-old children can understand and duly describe the pretend meanings enacted by play partner.However, against Leslie's(1987)Decoupling Theory, Harris(1994)claimed that children of this age don't understand other's mental attitudes.He insisted that children can participate in collaborative play without understanding other player's mind, because they can stipulate make-believe entities and understand them in their own mind.If this claim is true, we can infer that children of this age can't discriminate between the two false acts, namely that of pretend and that of false belief, which is accordance with Perner et al's(1994)hypothesis.In line with the paradigm of imagination-reality distinction, against Bretherton's(1989)claim, Lillard(1994)insisted that "at least by three years of age children appear to have a good grasp of the fact that the pretend world is separate and different from the real world, and that the pretend representation is defferent from reality".Hence, 3-year-old children can duly conduct the pretend play.On the other hand, against Leslie's(1987)Decoupling Theory, Lillard, A.proposed the Action Theory of Pretend(to be exact, the Action Referencing Theory of Pretend).This theory suggested the uniqueness of pretence among various kinds of "mental states":thinking, dreaming, wanting etc.Which is by large accordance with the paradigm of acting as if.However, she also examined the "concept of pretence"rather than "pretend action".Can we not apply the Prelief hypothesis to 2 and 3-year-old children in more action like context? To discuss and consider this issue, the experiment 2 of kayo(1992)was brought up, which investigated the children's pretend play by use of an intrusion method:In conclusion, for the further researches to investigate the developmental transition of this age, we should take the action wavering(between pre-pretend and pre-non pretend)theory rather than the propositional theory of acting as if.We may be able to apply this theory to autistic children to find their lack of flexibility of the wavering.