In this paper, we will show that eyebrow region can be used as a stand-alone biometric for recognition. We compare the recognition performance of eyebrow with eye and full face following NIST's FRGC Experiment 4 protocol, the most harsh experiment in FRGC ver2.0 database which involves matching 8,014 uncontrolled probe images from 466 subjects to 16,028 controlled target images from the same 466 subjects (∼128 million comparisons). We are the first to evaluate the recognition performance of eyebrow on such a big database and compare to the performance of full face. We will use multiple discrete transform encoded features to capture the local information of eyebrow as well as full face. The experimental results amazingly show that compared with the full face, the eyebrow region has 5/6 drop in size but only 1/6 drop in rank-1 identification rate.