The col-cap concept encouraged neurologists to inject a large group of muscles in the treatment of cervical dystonia. This includes deep muscles such as the obliquus capitis inferior or the semispinalis capitis, and muscles close to vascular or neurological structures such as scalene muscles. Our aim was to determine the accuracy of injections in cervical muscles using ultrasonography (US) or palpation of anatomical landmarks.A mix of paint, gelatin and iodized contrast agent was injected in nine pairs of cervical muscles of human cadavers, according to two injection techniques: US-guided and non-guided. The dye was localized on 1 cm-thick, frozen slices.A total of 102 muscles was injected in the US-guided group (n = 8). The global accuracy was 88.2%. The lowest accuracy was in the OCI (41.7%); trying to avoid the vertebral artery, injections were too medial. A total of 54 muscles was injected in the non-guided group (n = 3). The global accuracy was 48.0%; moreover, some dye was found in four blood vessels. The embalming process produced texture changes, making difficult the palpation of bony landmarks.Our results indicate that US-guided injections are more accurate than non-guided injections in most cervical muscles.