Four models of working memory processes in the complex span paradigm were tested: The task-switching model of Towse, Hitch, and Hutton (1998), the interference account of Saito and Miyake (2004), and two versions of the time-based resource-sharing model of Barrouillet, Bernardin, and Camos (2004). On the basis of a reading span paradigm that used segmented sentences, the effect of processing time on the recall of words was investigated while the amount of processing was held constant. Two conditions of reading (continuous vs. normal) were compared in order to study the influence of brief pauses during reading that could be used for articulatory rehearsal. The results favor a version of the time-based resource-sharing model: A faster reading rate had a negative effect on recall. The effect of reading rate was obtained with continuous as well as normal reading, revealing that even continuous articulation does not prevent simultaneous refreshing of memory traces. A second experiment showed that continuous reading made concurrent articulatory rehearsal virtually impossible. These findings imply that a second rehearsal mechanism for verbal working memory, other than articulatory rehearsal, exists.