Visual masking procedures are considered to have great potential for studying information processing that occurs outside of consciousness. Unfortunately, effects that indicate processing of masked word stimuli have been both difficult to obtain and, once obtained, difficult to replicate. The present seven experiments failed to obtain an effect of lexicality (word vs. nonword targets) on detection that was recently reported by Doyle and Leach (1988). Whereas Doyle and Leach had used backward binocular masking, most of the present experiments used simultaneous dichoptic masking. Doyle (1990) recently suggested that the effect of lexicality on detection (coupled with an effect of knowledge of results, which was also not obtained in the present research) could explain why Greenwald, Klinger, and Liu (1989) found no evidence for detectability of masked words that were nevertheless analyzed semantically. The differences of the present findings from those of Doyle and Leach (1988) not only confirm the uncertainty of generalizing across masking procedures, but also indicate that Greenwald et al.'s "detectionless processing" interpretation remains viable.