The primary aim of this study was to design and then test the intrarater reliability of a torque-controlled method of measuring passive abduction of the thumb carpometacarpal (CMC) joint. A secondary aim was to quantify passive CMC abduction in patients with and without contracture. Initially, clinicians used subjective feel (without range of motion measurements) to identify 52 people with loss of passive thumb CMC abduction. All subjects had a neurological condition. Passive thumb CMC abduction was measured in both hands of these 52 people and the hands of another 20 healthy able-bodied individuals (total of 72 people and 144 hands). Passive thumb CMC abduction was measured using a newly designed torque-controlled device and the previously recommended caliper method. Repeat measurements were taken with both devices, two to three days later, by blinded assessors on a subgroup of 12 patients (24 hands). Median (interquartile range) CMC angle of thumbs deemed by clinicians to have contracture was 45° (41–52°) and that of subjects without contractures was 56° (53–60°). The intraclass correlation coefficient for the repeat measures attained with the torque-controlled device was 0.78 (95% confidence interval, 0.56–0.90). The torque-controlled device provides a way of standardizing torque when measuring passive thumb CMC abduction. The clear difference between passive CMC abduction of subjects with and without contracture confirms the ability of clinicians to use feel and subjective assessment to identify patients with contracture.