This paper investigates vulnerability detection performance between hardware-based (NetClarity) and software-based (Nessus) test tools. However, both test tools use different standard to classify risk level. Not only that, vulnerability detection has weakness. It is unable to address the overall risk level of the scanned network. Therefore, this paper proposed“Network Risk Metric” to compensate overall risk evaluation from various vulnerability
detection tools. Two major ideas are applied: “Weight-Cut-off Severity Normalization” and “Probability of Trust”. Experimental results showthat the total estimated risk(%)derived from “Network Risk Metric” iseffective. It compromises the differences of risk detected from various test tools and can be used to represent the overall risk in more reasonable way. Fuzzy Logic will be applied to the “Network Risk Metric” in the future work.