Groups of Grad 1, high school students, and elderly subjects were tested for their ability to discriminate tactile forms. The performance of control subjects who received no prior training was compared with that of subjects who received prior visual training under normal viewing conditions (Vw), or prior visual training with special goggles restricting form perception to a series of part impressions (Vp). Both the visual and tactile versions of the task involved comparisons of standard figures matched either with themselves or with transformation figures. There was no evidence of transfer for the children, regardless of previous experience. Specific as well as nonspecific cross-modal transfer was found in the high school and elderly groups following Vp training. Positive transfer was also found for the high school students in the Vw condition, although there is some question whether this was the result solely of cross-modal effects. Whereas all age groups performed equally well on standard versus transformation comparisons, standard versus standard matches were highest in the high school students, lowest in the children, and intermediate in the elderly. The generally poor performance of the children and the inability on the part of the elderly to benefit from Vw training are discussed in terms of task difficulty and inefficient perceptual processing strategies.