Two experiments investigated the influence of perceptual fluency on recognition memory. Words were studied using a shallow encoding task to decrease the contribution of recollection on recognition. Fluency was manipulated by blurring half of the test probes. Clarity varied randomly across trials in one experiment and was grouped into two blocks (clear and blurry) in the other experiment. Clarity did not influence recognition judgments or the ERP correlate of familiarity (FN400) when clarity was blocked across trials, but fluent probes (old and clear) elicited a more negative ERP than less fluent probes 280-400 ms at parietal electrode sites. Random variations in clarity produced the opposite pattern of results because recognition judgments and FN400 amplitudes varied, whereas the early ERPs did not differ. The results are interpreted as evidence that blocking clarity across trials led to recognition that was based on repetition fluency differences (i.e., old more fluent than new), which was associated with an early (280-400 ms) ERP at parietal electrodes in the absence of FN400 differences. Randomly varying clarity across trials created a situation where repetition fluency and perceptual fluency (i.e., probe clarity) interacted and led recognition to be based on familiarity/conceptual implicit memory that was associated with FN400 amplitudes in the absence of early ERP differences. The behavioral and ERP differences suggest that perceptual fluency, by itself, is capable of supporting recognition in some contexts and that, in other contexts, fluency can combine with other memory trace information to support recognition. (Contains 1 table and 6 figures.)