In two completely randomized experiments, subjects were required to judge either which was the rougher of two abrasive papers or whether two abrasive papers were the same or different. Judgments were made visually, tactually, or with both vision and touch available. The subjects used either the right hand or the left hand in the touch conditions. Differences between the hands in terms of either proportion correct or mean latency were negligible in both experiments. Accuracy was statistically equivalent across conditions, although the latency of visual judgments was shorter. In the same-different experiment, comparable accuracy for vision and touch appeared to result from different strategies. Subjects in the touch condition were much less likely to be correct without guessing on “different” trials. In a third, within-subject experiment, a comparison was made of four probability models of dual-mode efficiency. Subjects appeared not to treat the two sources of information as independent; rather, the probability of a correct response in the combined vision-touch condition could be best described as the arithmetic mean of the vision and touch conditions. Latencies for the combined condition also appeared to reflect a similar compromise. Implications for further research are discussed.