Functional magnetic resonance imaging (fMRI) is one of the techniques capable of detecting human brain function noninvasively. Although many fMRI studies have adopted echo-time (TE) of 30 to 40ms aiming to maximize blood oxygen level-dependent (BOLD) signal change, large distortion and signal dropout occur near air-filled spaces such as frontal sinus in functional imaging because of the magnetic field irregularities. Theoretically, shorter TE could decrease the magnetic distortion and signal dropout near air. However, it has not been determined whether short TE of less than 30ms may affect the statistical t value of functional image, which is the most important factor to evaluate results of fMRI study. In the present study, we assessed whether shorter TE can be utilized for fMRI study of the human brain. Nineteen healthy volunteers were scanned using fMRI during finger tapping and viewing checker-circle tasks. Signal dropout near the sinus, statistical t value, percent signal change, and modified temporal signal to noise ratio (mtSNR) of functional images were evaluated in the primary motor cortex and primary visual cortex while TE was varied between 20 and 35ms. Signal dropout was the smallest at a TE of 20ms, whereas there was no significant effect of TE on the statistical t value both in the primary motor cortex and in the primary visual cortex activity. These findings suggest that 20ms is the optimal TE value in the range of 20ms to 35ms for fMRI.