Abstract. Client-based attacks on internet users with malicious web pages represent a serious and rising threat. Internet Browsers with enabled active content technologies such as JavaScript are vulnerable to so-called drive-by downloads. Drive-by downloads are able to automatically infect a victim's system during a single visit of a crafted web page testing various vulnerabilities and installing e.g. malware files or illegal content without user interaction. In this paper we present MonkeyWrench, a low-interaction web-honeyclient allowing automatic identification of malicious web pages by performing static analysis of the HTML-objects in a web page as well as dynamic analysis of scripts by execution in an emulated browser environment. Using this hybrid approach MonkeyWrench overcomes shortcomings of existing low-interaction web-honeyclients in dealing with obfuscated JavaScript while outperforming high-interaction systems. Further MonkeyWrench is able to identify the exact vulner...