We assessed the length-tension relationship of the posterior deltoid to triceps transfer in 8 tetraplegics (n = 11 transfers) and compared the results to the length-tension relationship of the normal triceps measured in a check sample composed of 9 able bodied, right handed women. We designed a device to lock the arm and forearm and used a force transducer to assess the torque output isometrically. The muscle was tested at 6 different lengths (130, 110, 90, 70, 45 and 0° of elbow flexion) with the shoulder abducted at 90°. As expected, the transfer behaved differently from the normal triceps. The mean maximum torque recorded was 7. 8 Nm in patients while it was 27 Nm in the check sample. When compared, the absolute values (ie values expressed with a dimension of torque) were significantly different between groups (0. 00001 < P < 0. 002). The expression of this relation (ie the relative values expressed as percentage of maximum values) revealed significant statistical differences (p < 0. 002) at 90 and 70° of elbow flexion; the peak torque was recorded at 130° in patients while it was recorded at 110° in the check sample, with a plateau between 110° and 70°. On the other hand, if the length-tension relationship was fairly similar among subjects of the check sample, it exhibited tremendous differences among patients; it seemed that initial tension given by the surgeon represented a variable difficult to control without a device dedicated to that task.