Two eye-movement experiments are reported in which a boundary paradigm was used to manipulate the presence versus absence of boundaries for high-frequency and low-frequency target words in the parafovea. In Experiment 1, this was done by introducing a blank space after the target words, whereas in Experiment 2 this was done by rendering the target words in red. In both experiments, higher frequency targets engendered longer saccades, whereas the presence of parafoveal word boundaries engendered shorter saccades. This pattern suggests the operation of two countermanding saccade-targeting mechanisms: one that uses parafoveal processing difficulty to adjust saccade lengths and a second that uses word boundaries to direct the eyes toward specific saccade targets. The implications of these findings for models of eye-movement control during reading are discussed, as are suggestions for integrating dynamic-adjustment and default-targeting accounts. Keywords Chinese reading. Dynamic-adjustment model. Eye-movement control. Word segmentation Skilled reading requires the coordination of lexical processing with eye movements to rapidly extract visual information from the printed page. One important aspect of this coordination is saccade targeting, or the Bdecisions^about where to move the eyes. Because much of what has been learned about eyemovement control in reading has come from studies involving alphabetic writing systems, models of how readers decide where to move their eyes have been informed largely by what has been learned from studying languages like English and German, where the eyes seem to be directed toward a small number of default targets (e.g., see Engbert, Nuthmann, Ricter, & Kliegl, 2005; Reichle, Pollatsek, & Rayner, 2012). For example, with English, readers seem to direct their eyes slightly to the left of a word's center, toward the preferred-viewing location (PVL;