Meyer, Schvaneveldt, and Ruddy (1972) reported that in a lexical decision task, the word association effect was maintained when two associated words were separated by an unassociated word but was eliminated when the two were separated by a nonword. They suggest that the effect is eliminated in the latter case because a negative decision may result in a “resetting” of the memory system to a neutral level. The present investigation comprised a partial replication of the above study. Associated words were separated either by unassociated words, pronounceable non words, or unpronounceable non words. Results indicated that the word association effect was significant regardless of whether the intervening item was a word or a nonword; and, in addition, the pronounceability of the nonword was not found to influence the magnitude of the effect. The data were taken to provide further support for the spreading excitation model.