Although lameness is a common problem in elephants (Asian elephant [Elephas maximus] and African elephants Loxodonta africana and Loxodonta cyclotis) under human care, there has not been a standardized lameness assessment system to date. This study developed and evaluated a standardized system for the assessment of locomotion in elephants under human care regardless of husbandry system. In total, 72 elephants out of a possible 73 in the United Kingdom and Ireland were filmed from behind, from in front, and from both sides. Using a questionnaire and a select panel of elephant specialists, a zoo veterinarian, and a locomotion expert, a numerical rating scoring (NRS) system was proposed. Locomotion was scored on a 4-point scale with numerical values 0–4 corresponding to specific criteria as follows: 0 = clinically sound, 1 = stiffness, 2 = abnormal tracking, and 4 = reluctance to bear weight. The intra- and interobserver repeatability of five veterinary surgeons using this system was determined and compared with a visual analog scale (VAS) expressed as a 100-mm line. Overall intraobserver reliability was moderate (Cohen's kappa [κ] = 0.676) and interobserver reliability was fair (κ = 0.37) for the presence of lameness. Interobserver agreement improved from the first scoring to second scoring from slight agreement to fair agreement for stiffness and reluctance to bear weight. Abnormal tracking had moderate intraobserver agreement for both scoring sessions. There were wide widths of agreement for the VAS interobserver (67 mm); however, they were narrower for the intraobserver (33 mm). The developed NRS can be used on freely moving elephants to evaluate elephant locomotion, regardless of husbandry methods, and has been shown to be more reliable than a VAS.