In unconscious semantic priming, an unidentifiable visually masked word (th
e prime) facilitates semantic classification of a following visible related
word (the target). Three experiments reported here provide evidence that m
asked primes are analyzed mainly at the level of word parts, not whole-word
meaning. In Experiment I, masked nonword primes composed of subword fragme
nts of earlier-viewed targets functioned as effective evaluative primes. (F
or example, after repeated classification of the targets angel and warm, th
e nonword anrm acted as an evaluatively positive masked prime.) Experiment
2 showed that this part-word processing was potent enough to oppose analysi
s at the whole-word level. Thus, smile functioned as an evaluatively negati
ve (!) masked prime after repeated classification of smut and bile. Experim
ent 3 found no priming when masked word primes contained no parts of earlie
r targets. These results suggest that robust unconscious priming (a) is dri
ven by analysis of part-word information and (b) requires previous classifi
cation of visible targets that contain the fragments Inter serving as prime
s. Contrary to a widely held view, analysis of subliminal primes appears no
t to function at the level of analysis of complete words.