Rhesus monkeys were trained on a within-subjects design to assess whether they could perform concurrently visual-to-haptic (V-H) and haptic-to-visual (H-V) crossmodal delayed matching-to-sample (DMS). A parametric analysis was conducted of the effect of delay between presentation and re-presentation of the test discriminanda (three-dimensional geometric objects). The results indicate that (a) monkeys are capable of concurrent V-H and H-V crossmodal matching of objects by shape, size, and texture; (b) monkeys acquire faster and perform better crossmodal matching in the V-H direction than in the H-V direction; (c) as they learn to perform DMS with successive object pairs, monkeys transfer some--procedural--knowledge from the use of one pair to the use of the next; and (d) in the monkey, crossmodal short-term memory, as measured by DMS performance, has a temporal decline.