Visual processes in Grade 1 were examined for their predictive influences in nonalphanumeric and alphanumeric rapid naming (RAN) in 51 poor early and 69 typical readers. In a lagged design, children were followed longitudinally from Grade 1 to Grade 3 over 5 testing occasions. RAN outcomes in early Grade 2 were predicted by speeded and nonspeeded visual processing measures, after controlling for initial (Grade 1) RAN, matrix reasoning, phonological awareness, and word decoding abilities. A predictive influence of backward visual masking--a speeded visual discrimination task--was found for nonalphanumeric RAN in early Grade 2 but not for alphanumeric RAN or subsequent RAN ability in Grades 2 and 3. A nonspeeded predictor involving controlled visual attention accounted for significant variance in early Grade 2 RAN in the poor early reader group. Results are discussed in relation to Wolf, Bowers, and Biddle's conceptualization of rapid naming--in particular, on the roles of visual processes in speeded low and nonspeeded high spatial frequency visual information in predicting RAN.