A previous study by Gibson et al. (1972) that investigated the effect of orthographic structure on letter search was replicated and extended in order to identify factors that might explain the apparent discrepancy between their results and those of comparable studies. Experiment one tested whether the discrepancy might be explained by difference in (1) the sensitivity of a between-subjects design as opposed to a within-subjects design, or (2) the overall range of orthographic structure represented in the displays, when subjects have been equated in terms of their rate of search through random strings of consonants. Experiment two was concerned with the psychological reality of the differences in the orthographic structure of the letter strings presented in experiment one. In comparing the results of the two analyses of the search time data, it might be argued that the negative findings of the Gibson et al. study may be explained by the nature of the constraints they adopted in constructing their orthographically well-structured and poorly structured strings. It is suggested that unlike the more traditional, between-subjects design, a within-subjects design permits subjects to compare differences in orthographic structure. This encourages them to adopt more efficient search strategies for strings having more than a minimum amount of structure. (SW)