Orthographic priming was examined using the masked priming paradigm combined with event-related potential (ERP) recordings. In Experiment 1, low-frequency target words were primed either by the same word, a high-frequency orthographic neighbor, or an unrelated word. In Experiment 2, effects of high-frequency orthographic neighbor primes were compared with effects of nonword neighbor primes. High-frequency neighbor primes were found to generate a different priming effect in the ERP waveform, as compared with both repetition primes and nonword neighbor primes. This differential priming effect was mostly evident in the P325 (starting at around 300 msec post-target-onset) and N400 components. In these epochs, word neighbor primes produced greater negativities than did unrelated primes, whereas full-repetition primes and nonword neighbor primes produced less negativity than did unrelated primes. The results help locate the influence of high-frequency neighbor primes at the level of computations operating on word-form representations and the mapping of word forms onto meaning.