The objective of creating malicious software (i.e., malware), intruding computers and conducting malicious activities is shifted from showing off attacker's computer skills to earning money. Thus, recent attackers take more sophisticated and effective malware infection ways such as malware infection via malicious Web sites as well as the traditional exploitations like worm propagation. The malicious Web sites attempt to compromise machines by drive-by-download attack which redirects users to exploiting sites and install malware compulsorily in their machines by exploiting vulnerabilities of their Web browser or plugins. As a countermeasure for these malicious Web sites, blacklisting URLs or domains of them is significant. However, attackers tend to change the URLs or domains in a short period to avoid the blacklist. Thus, a blacklisting scheme which can filter even unknown malicious Web sites is critical. In this paper, we first analyze characteristics of malicious Web sites by their domain information such as AS (Autonomous System), IP address block, IP address, domain, and registrar. Second, we evaluate reputations of IP address blocks and registrars used by attackers. Then, we propose a blacklisting scheme constructed of the combination of IP address block and registrars with low reputation, that is, intensively used by attackers. From our experimental results, the Web sites with the same combination with low reputation appeared over long period, which indicates that our proposed blacklist has a certain capability of filtering unknown malicious Web sites.