Adults show a deficit in their ability to localize tactile stimuli to their hands when their arms are in the less familiar, crossed posture (e.g., Overvliet et al., ; Shore et al., ). It is thought that this ‘crossed-hands effect’ arises due to conflict (when the hands are crossed) between the anatomical and external frames of reference within which touches can be perceived. Pagel et al. () studied this effect in young children and observed that the crossed-hands effect first emerges after 5.5-years. In their task, children were asked to judge the temporal order of stimuli presented across their hands in quick succession. Here, we present the findings of a simpler task in which children were asked to localize a single vibrotactile stimulus presented to either hand. We also compared the effect of posture under conditions in which children either did, or did not, have visual information about current hand posture. With this method, we observed a crossed-hands effect in the youngest age-group testable; 4-year-olds. We conclude that young children localize tactile stimuli with respect to an external frame of reference from early in childhood or before (cf. Bremner et al., ). Additionally, when visual information about posture was made available, 4- to 5-year-olds’ tactile localization accuracy in the uncrossed-hands posture deteriorated and the crossed-hands effect disappeared. We discuss these findings with respect to visual–tactile-proprioceptive integration abilities of young children and examine potential sources of the discrepancies between our findings and those of Pagel et al. ().