A large body of research now supports the claim that two different and dissociable processes are involved in making numerosity judgments regarding visual stimuli: subitising (fast and nearly errorless) for up to 4 stimuli, and counting (slow and error-prone) when more than 4 stimuli are presented. We studied tactile numerosity judgments for combinations of 1-7 vibrotactile stimuli presented simultaneously over the body surface. In experiment 1, the stimuli were presented once, while in experiment 2 conditions of single presentation and repeated presentation of the stimulus were compared. Neither experiment provided any evidence for a discontinuity in the slope of either the RT or error data suggesting that subitisation does not occur for tactile stimuli. By systematically varying the intensity of the vibrotactile stimuli in experiment 3, we were able to demonstrate that participants were not simply using the 'global intensity' of the whole tactile display to make their tactile numerosity judgments, but were, instead, using information concerning the number of tactors activated. The results of the three experiments reported here are discussed in relation to current theories of counting and subitising, and potential implications for the design of tactile user interfaces are highlighted.