The Serbian writing system was used to investigate whether a serial procedure is implicated in print-to-sound translation and whether components of the reading aloud system can be strategically controlled. In mixed- and pure-alphabet lists, participants read aloud phonologically bivalent words comprising bivalent letters in initial or final positions. Words with bivalent letters in initial positions were disadvantaged relative to nonbivalent controls to a greater degree than were words with bivalent letters in final positions, and the size of the effect was greater in the mixed-alphabet situations than it was in the pure-alphabet situations. A dual-route theory of bialphabetic reading aloud is proposed in which the nonlexical procedure operates serially and nonlexical spelling-sound correspondences for each script can be strategically emphasized or deemphasized.