According to many researchers, implicit learning in the serial reaction-time task is predominantly motor based and therefore should be independent of stimulus modality. Previous research on the task, however, has focused almost completely on the visual domain. Here we investigated sequence learning when the imperative stimuli were presented tactilely to the fingers. Learning in this task was compared to sequence learning in a typical visual task, using very similar experimental conditions. The results indicate that sequential learning occurs in the tactile task, though to a lesser degree than in its visual counterpart. Furthermore, there was similar cross-modal transfer in both directions, meaning that transfer from the visual to the tactile task was partial. It is proposed that sequence learning involves a stimulus-specific component in the visual but not in the tactile task.