Chemodenervation of cervical muscles with botulinum A toxin, although useful in treating spasmodic torticollis, has been associated with dysphagia. Retrospective analysis of dose and injection site (sternomastoid vs. posterior cervical muscle groups) in 26 patients (49 injections) suggested that dysphagia was related to the quantity of toxin injected into the sternomastoid muscle: dysphagia, median 150 IU (7 injections); and no dysphagia, median 100 IU (42 injections; p = 0.026 Wilcoxon test). In a prospective study (31 injections to 24 patients), limiting the dose administered to the sternomastoid to 100 IU, substantially reduced the incidence of dysphagia (0 of 31, p = 0.27, Fisher's exact test). Denervation of human orbicularis muscle fibers, 5 weeks to 4 months after injection of botulinum A toxin for the treatment of blepharospasm, was successfully demonstrated by intense, diffuse acetylcholinesterase staining. A weight-adjusted dose similar to that given for torticollis was injected into longissimus dorsi muscle in 6 albino rabbits. Using the acetylcholinesterase stain as a marker, a diffusion gradient was noted over a distance of 30 to 45 mm from the point of injection and in contralateral muscle 15 to 25 mm from this point. Thus, denervation was demonstrated to occur within a definable area which crossed anatomic barriers, such as fascia and bone. These clinical and laboratory data suggest that dysphagia following botulinum toxin therapy results from toxin spread to pharyngeal musculature from the sternocleidomastoid injection site. Limiting of the injection dose to 100 IU or less to the sternomastoid substantially decreases the incidence of this complication.