The purpose of this study was to replicate the structure of mental speed and relations evidenced with fluid intelligence (Gf) found in a number of recent studies. Specifically, a battery of computerized tasks examined whether results with paper-and-pencil assessments held across different test media. Participants (N = 186) completed the battery, which incorporated 20 elementary cognitive tasks, 4 broad speediness (Gs) measures, and 5 Gf markers. Competing measurement models were tested. A higher-order model, with a general mental speed factor and 7 task-class specific factors fit the data well. Gs could not be distinguished from general mental speed. Besides the general mental speed factor, two task-class specific factors were moderately related to Gf. These findings strengthen the evidence for a multifacted structure of mental speed, and highlight the importance of specific speed task-classes in accounting for meaningful outcomes. (Contains 4 tables and 2 figures.)