This research had been examining the effects of three factorslevel--of--processing (gtaphemic-semantic), test instruction (implicit-explicit) and the type of test (word--stem and word-fragment)- on priming in memory tests. On the basis of transfer- appropriate- processing (TAP) model and retrieval intentionality criterion (RIC), and with respect to the characteristics of word stem and word fragment tests, it was predicted that the subjects, in the condition of implicit test instruction, do differently in word stem and fragment tasks. The subjects were 80 university undergraduates, randomely assigned in 8 experimental groups. As expected,each of the three factors had significant effect on priming specifically in implicit condition, the tests were not similar: level of processing affected the word fragment, but not the stem. In addition, on the basis of these findings, it can be concluded that word fragment task, is not a pure perceptual implicit test, and word stem under, explicit test instruction, is a cued- recall test, and under implicit test instruction, is a perceptual implicity one.