It has been reported (Hamburger, 1980) that tactile point localization (locognosia) is more precise in the transverse than longitudinal axis of the arm. However, a potential motor-sensory confound was inherent to the technique employed which involved the subject manually marking the perceived stimulus site. We have re-examined the issue using a response mode that did not depend upon the subject’s movement accuracy and have extended observations to the lower limb. Twenty-eight (19 female, 9 male, mean ± S.D. age 20.6 ± 0.7 years) healthy, right-handed subjects participated with local ethics committee approval. A 7 x 7 cross of stimulus points (13 points, 5 mm separation), arrayed along intersecting lines, orientated in the transverse and longitudinal limb axes, was drawn on the shaved skin of the left forearm and lower leg. In each trial, a brief tactile stimulus was first applied, with a von Frey hair (rating 150 mN), to the central locus of the cross (reference), followed, after 1s, by a stimulus to one of the 13 test loci. The limb was obscured from the subject’s view. The subject was required to state whether the test locus was ‘more distal’/ ‘more proximal’ (longitudinal axis) or ‘more lateral’/’more medial’ (transverse axis) to the reference locus. The four limb-axis combinations were tested separately in each subject. Each test locus received 10 stimuli (total 70 stimuli per axis). For each subject, at each test locus, the probability of the response ‘more distal’ (longitudinal axis) or ‘more lateral’ (transverse axis) was calculated. The interval of uncertainty (IU, a measure of discriminatory threshold) and point of subjective equality (PSE, the test locus perceived to correspond to the reference locus) were estimated from psychophysical functions (probability versus stimulus locus). The group mean ± S.D. values (in mm) of IU for the different limb-axis combinations were: arm-transverse, 0.96 ± 0.11; arm-longitudinal, 1.72 ± 0.65; leg-transverse 1.17 ± 0.35; leg-longitudinal, 1.82 ± 0.51. Group mean PSE values were consistently small (< 1mm). Statistical analysis (2-way, repeated measures ANOVA) of IU values indicated significant main effects of axis (p = 0.000) and limb (p = 0.045), whilst the axis*limb interaction was non-significant (p = 0.614). Thus, our subjects showed (1) greater locognostic acuity (smaller IU) in the transverse than longitudinal axis of both the forearm, confirming Hamburger's (1980) findings, and lower leg and (2) more accurate tactile discrimination in the upper than lower limb.