The previous studies have focused mostly on the relationship between the human resources management systems and the firm performances. They have showed that human resource management (HRM) system is positively related to organizational performance. However, only few studies have attempted to find out various types of human resource management system. Since the previous studies have classified the types of HRM systems according to the partial and fragmented criteria, more integrated approaches are in needs to be taken to classify HRM systems. Furthermore, not many studies trying to prove the relationship between the various types of HRM systems and the various types of firm performance (i.e., turnover rate, innovative performance, and financial performance) have been done so far. The purpose of this study is to classify firms according to theoretically identified HRM systems, commitment type, star type, paternalistic type, and control type. It tries to show how the different types of HRM systems are related to various types of firm performances. To classify HRM systems, this study employed the approach with two dimensions - ⅰ) the mode of human existence in an organization(individual-relational continuum), and ⅱ) the control mechanism(authority-autonomy continuum) -based on the previous studies. This study used the data from the Workplace Pannel Survey(WPS) 2005 of the Korea Labor Institute. The WPS 2005 received the responses from managers in charge of the human resource departments at 1,905 Korean establishments. Out of 1,905, the 931 samples from manufacturing and service industry, and union and nonunion firms with more than 50 employees were used in this study. To find out the characteristics of HRM systems, this study used two-step cluster analysis in SPSS, and to compare the level of organizational performance among these HRM systems ANOVA and MANCOVA were used. The empirical results showed that HRM systems were classified into commitment type, star type, paternalistic type and control type as expected. We also found out the levels of organizational performance were significantly different among these four types of HRM systems. Both commitment type and paternalistic type were associated with lower levels of turnover rate than star and control type. Innovative performance was highest in the commitment type, next highest with star type and next highest with paternalistic type. Innovative performance was lowest under control type. In terms of financial performance, both commitment type and star type were superior to the other types of HRM system. Financial performance was lowest under control type. From these results, we concluded that most of our hypotheses were supported. We also found that categorical variables composed of HRM policies and practices were very useful for the analysis of HRM classifications. These findings help us to enrich the understanding of HRM systems and provide valuable practical implications to HRM managers. This study empirically proved the theoretical foundation in that the firms with the commitment type of HRM system achieve better performance than the other types of HRM system. The commitment HRM system influences more positively firm performance than the others since it stresses both relational and autonomous orientation. Secondly, this study made a step forward methodologically. The two step clustering technique used in this study included both continuous and categorical variables, which made it possible to classify HRM systems effectively. Furthermore, with the HOMALS this study discussed how a particular HRM system is related to individual HR practices. The commitment HRM system, for example, has special features such as small group activities, suggestion program, multi-skill training, and higher wage level. The star system has pay system based not on seniority but on competence, performance related weeding-out and pay differentiation system, merit pay, and exceptional promotion. The paternalistic system retains seniority based pay and non differentiated wage level.