In this paper, we propose three metrics for detecting botnets through analyzing their behavior. Our social infrastructure (i.e., the Internet) is currently experiencing the danger of bots? malicious activities as the scale of botnets increases. Although it is imperative to detect botnet to help protect computers from attacks, effective metrics for botnet detection have not been adequately researched. In this work we measure enormous amounts of traffic passing through the Asian Internet Interconnection Initiatives (AIII) infrastructure. To validate the effectiveness of our proposed metrics, we analyze measured traffic in three experiments. The experimental results reveal that our metrics are applicable for detecting botnets, but further research is needed to refine their performance.