The current study investigates the validity of the visual scanning hypothesis, which posits that rapid automatized naming (RAN) predicts reading skill partly because both require the ability to perform rapid sequential eye-movements. Our data consist of eye-movements collected while 124 young English speaking adults of variable reading skill read passages and performed six modifications of RAN. These modifications isolated articulatory, lexical, oculomotor and attentional task components of RAN. A further requirement for participants was to perform each of the RAN tasks in two directions--the habitual direction of reading (RAN forward) and from right to left and top to bottom (RAN backward). Participants who were better at oculomotor control in RAN-like tasks were better readers regardless of task type or direction. Our most crucial finding is that the explanatory contribution of oculomotor control in the RAN-reading relationship is independent of the practice effect afforded by the habitual direction of visual scanning in reading.