The objectives of this study were to evaluate the likelihood of successful arthrocentesis of the equine elbow joint using the caudolateral approach and to determine if the deep branch of the radial nerve (DBRN) varies in its proximity to the site of centesis. Methylene blue (MB) was injected into 71 elbow joint specimens immediately caudal to the lateral collateral ligament using a 3.8-cm needle advanced to its hub. The elbow joints were dissected, staining of the synovial structures assessed and the proximity of DBRN to the site of centesis evaluated. The articular cartilage of all 71 joints was stained with MB. The location of DBRN did not vary substantially among the specimens and did not course close to the site of centesis. Direct communication was found between the bursa of the tendon of the ulnaris lateralis muscle and the elbow joint in 41 of 71 specimens (57.8 per cent). The caudolateral approach for centesis of the equine elbow joint, performed by inserting a needle 3.8-cm, was found to be reliable. Radial nerve paralysis reported to be caused by injection of local anaesthetic solution using the caudolateral approach may be due to diffusion of the solution from the dorsal pouch rather than from leakage at the site of centesis.