Background The estimation of relative distance is a perceptual task used extensively in everyday life. This important skill suffers from biases that may be more pronounced when estimation is based on haptics. This is especially true for the blind and visually impaired, for which haptic estimation of distances is paramount but not systematically trained. We investigated whether a programmable tactile display, used autonomously, can improve distance discrimination ability in blind and severely visually impaired youngsters between 7 and 22 years-old. Methods Training consisted of four weekly sessions in which participants were asked to haptically find, on the programmable tactile display, the pairs of squares which were separated by the shortest and longest distance in tactile images with multiple squares. A battery of haptic tests with raised-line drawings was administered before and after training, and scores were compared to those of a control group that did only the haptic battery, without doing the distance discrimination training on the tactile display. Results Both blind and severely impaired youngsters became more accurate and faster at the task during training. In haptic battery results, blind and severely impaired youngsters who used the programmable display improved in three and two tests, respectively. In contrast, in the control groups, the blind control group improved in only one test, and the severely visually impaired in no tests. Conclusions Distance discrimination skills can be trained equally well in both blind and severely impaired participants. More importantly, autonomous training with the programmable tactile display had generalized effects beyond the trained task. Participants improved not only in the size discrimination test but also in memory span tests. Our study shows that tactile stimulation training that requires minimal human assistance can effectively improve generic spatial skills.