Most models predict that priming a word should retard recognition of another sharing its initial sounds. Available short lag priming data do not clearly support the prediction. The authors report 7 continuous lexical-decision experiments with 288 participants. With lags of 1-5 min between prime and probe, response time increased for a monosyllabic word preceded by a word sharing its onset and vowel (but not one sharing its rime) and for a polysyllabic word preceded by another sharing its first syllable. The effect was limited to words primed by words, suggesting that identifying the prime strengthens its lexical attractor, making identification of a lexical neighbor more difficult. With lags of only a few trials, facilitatory effects of phonological similarity or familiarity bias effects were also seen; this may explain why clear evidence for inhibitory priming has been lacking hitherto.