Davis’ Technology Acceptance Model(TAM) has been well received in MIS research since its inception. The popularity of TAM is based on the virtues such as parsimony, concrete theoretical supports and easy adaptability to various adoption contexts. Recently, however, there was criticism on the overuse of the model by modification and extension. This study is to investigate the limitations and future directions of TAM by performing a series of quantitative meta-analyses on the TAM studies in both Korea and US. Before performing the metaanalyses, the extant six studies of TAM meta-analysis are compared and reviewed critically. Especially, we discussed the problems of those six studies in dealing with the moderator exploration of TAM. The data for the meta-analyses are collected from 84 studies in 39 Korean articles and 40 US articles. Since the results show that all the relationships in TAM are statistically significant in both countries, we may say TAM is generally supported regardless of country. Some differences, however, are found that all of the correlation coefficients and standard deviations are greater in the Korean studies than the US ones, and that the Korean studies paid less attention on the relationship between intention and behavior than the US ones. Even importantly, all the correlation coefficients fail to pass the Q test, meaning that they are heterogeneous so that we need to explore moderators of the relationships. Before performing the analysis for exploring moderators, the classification schemes for culture, technology, and users are proposed and explained. For culture, national culture is selected because of its popularity in social sciences and distinct differences between Korea and US. For technology, four types of IT are proposed with two criteria, enjoyment and risk. For users, we examined both whether the samples are students or not and whether the samples are proxy or not. The moderator analysis shows that national culture, risk as a technology characteristic, and student as a sample characteristic are statistically significant. Especially, we find that there is an interaction effect between national culture and enjoyment, which confirms that both national culture and technology characteristics are multi-dimensional. Based on the results of all the analyses, some suggestions are presented. First, when using TAM in countries other than US, national culture should be considered in setting research models and interpreting results. Second, when applying TAM to newly developed technologies, technology characteristics should be reflected in constructing research models. Third, we should pay attention to the possible bias from student samples. Finally, the future TAM study should be devoted to the theoretical development of the IS field rather than simply enlarging the TAM application domain.