This study examined selective attention to tactile dimensions by combining a selective cueing paradigm with a test of integrality. In Experiment 1, subjects selectively attended to changes in the frequency or duration of pairs of vibrotactile stimuli and identified the higher frequency or longer duration stimulus. In Experiment 2, using surface gratings in an identical experimental procedure, subjects identified the rougher or longer duration stimulus. In both experiments, greater performance accuracy was found on trials where the cue correctly (valid) predicted the changing dimension, vs incorrectly (invalid) cued or no-cue (neutral) trials. More errors on the invalidly vs neutrally cued trials show the cost of focal attention. Increases in performance on validly vs neutrally cued trials show a benefit of filtering irrelevant stimuli in the cued conditions. Results effectively demonstrate focal attention to tactile features. Tests of integrality, in terms of the effects of correlated change in both dimensions, showed no redundancy gain for either vibrotactile or grating tasks, suggesting that frequency and roughness are separable from stimulus duration. Interference of negative correlated change for frequency but not roughness discriminations may be explained by differences in task difficulty.