We report an experiment that evaluates whether BLANK or an unrelated prime is a more suitable baseline for assessing priming for an ERP study. Sixteen subjects performed a lexical decision task with a 1 s prime-target stimulus onset asynchrony. Increased amplitude for the N400 was observed for targets in the unrelated prime condition whereas targets in the BLANK prime condition evoked activity that was more like that in the related prime condition. Theoretically, we conclude that the N400 reflects semantic integration. Pragmatically, we conclude that the BLANK prime is a better neutral prime but that unrelated primes yield stronger N400 effects.