This study determined the just noticeable difference, or discrimination threshold, for manual stiffness perception and the test-retest reliability of this measure. Elastic, non-biological stiffness stimuli within the range found in human spines were generated by a device incorporating metal springs. The method of constant stimuli was used to estimate the just discriminable change in stiffness and results were expressed as a percentage of the base stiffness, or Weber fraction. A total of 25 physiotherapists and lay people participated on two separate occasions. Reliability of repeated measurements was evaluated by intraclass correlation coefficients, percentage of agreement scores and the standard error of the measurement. All three reliability coefficients showed high/good reliability for the repeated measurements. The mean Weber fraction was 7.7%, implying high sensitivity to differences in non-biological stiffness. The stability of the Weber fraction means that this index can be used as a measure of a subject's sensitivity to stiffness stimuli. Consequently, training programmes and various manual techniques designed to enhance therapists' stiffness discrimination can be evaluated.