Do native readers segment polysyllabic words based on orthographic/morphological criteria or phonological criteria? Research by Taft (1979, 2001) argues in support of the former, as readers were faster in split-word lexical decision tasks when the words were segmented by orthographic/ morphological principles based on Basic Orthographic Syllable Structure (or BOSS) units than when they were phonologically segmented following the Maximum Onset Principle (MOP). However, a BOSS-based preference has been difficult to replicate. The present research examined three factors potentially modulating a BOSS-based segmentation preference: whether a given BOSS unit is or is not present in other words, reading experience, and word frequency. The results showed that across higher and lower ability readers, and across words with shared or unique BOSSes, a BOSS preference was reliably obtained in low but not in high frequency words. Thus, word frequency appears to modulate the segmentation strategy of polysyllabic English words. (Contains 4 tables and 6 footnotes.)