Two experiments on finger localization are reported. Experiment 1 compared children who were poor readers with two groups of children matched to poor readers for sex and either age (CA controls) or reading age (RA controls). The participant's hands were kept out of his or her sight in a semi-open box while the fingers of one hand were lightly touched by the experimenter. The participant's task was to indicate, using the thumb of either the same hand (within-hand condition) or the opposite hand (across-hands condition) to respond, the finger(s) which had been touched by the experimenter. Performance was significantly impaired in the across-hands condition compared with the within-hand condition. Experiment 2 was carried out with dyslexic adults and a control group of normal readers. Using the same method of responding as in Experiment 1, a significant deficit in the across-hands condition compared to the within-hand condition was found for both groups. The effect was also obtained for both groups when participants were required to point to the relevant fingers on a photograph of a hand rather than use the thumb of the opposite hand to respond. The results are discussed in relation to the hypothesis of a deficit in inter-hemispheric transfer of tactile information in dyslexia.