The majority of children and adults with reading disabilities exhibit pronounced difficulties on naming-speed measures such as tests of rapid automatized naming (RAN). RAN tasks require speeded naming of serially presented stimuli and share key characteristics with reading, but different versions of the RAN task vary in their sensitivity: The RAN letters task successfully predicts reading ability, whereas the RAN objects task does not reliably predict reading after kindergarten. In this study we used functional magnetic resonance imaging to evaluate the neural substrates that may underlie performance on these tasks. In two scans during the same test session, adult, average readers covertly rapidly named objects or letters or passively viewed a fixation matrix of plus signs. For both rapid naming tasks compared with fixation, activation was found in neural areas associated with eye movement control and attention as well as in a network of structures previously implicated in reading tasks. This reading network included inferior frontal cortex, temporo-parietal areas, and the ventral visual stream. Whereas the inferior frontal areas of the network were similarly activated for both letters and objects, activation in the posterior areas varied by task. The letters task caused greater activation in the angular gyrus, superior parietal lobule, and medial extrastriate areas, whereas object naming only preferentially activated an area of the fusiform gyrus. These results confirm that RAN tasks recruit a network of neural structures also involved in more complex reading tasks and suggest that the RAN letters task specifically pinpoints key components of this network.