The speed and accuracy with which subjects can read words is enhanced or “primed” by a prior presentation of the same or related words. The masked priming paradigm has shown that even when the subjects were not consciously aware of the briefly presented primes, priming effects could still be observed. However, it is not clear that whether the conscious identification of the primes will be required for the long-term priming effect, where the SOAs are longer than few seconds. In the current study, we used a mirror-word identification paradigm to investigate this question. Words in mirror-image orientation were presented on the computer screen, and subjects pressed a button as soon as they identified the word. Some words were presented once (control group), and some words were presented twice. The items were intermixed in pseudo-random order. The duration of the time for displaying the primes was adjusted to match each participant’s threshold of word identification so that they would identify half of the primes and miss the other half. We compared the reaction times for identifying the targets following these two conditions to their correspondent control groups to examine if they both have priming effect and their magnitude. Now we are also using event-related fMRI technique to investigate the brain activation regions, and also to test the hypothesis that the fMRI signal will increase while the primes are not identified instead of the signal reduction results that have been observed in most of the priming studies.