Background Dysphagia (swallowing impairment) is a common and often life‐threatening problem after stroke. Submental surface electromyography (ssEMG) visual biofeedback is a commonly used clinical tool to train novel swallowing maneuvers, even though its effectiveness has been questioned. Objective To compare the effect of ssEMG and videofluoroscopy (VF) visual biofeedback on swallowing airway protection accuracy when training the volitional laryngeal closure swallowing maneuver (vLVC) in poststroke patients with dysphagia. Researchers also examined whether clinicians accurately judged vLVC performance. The hypothesis was that patient vLVC accuracy and clinician verbal cue accuracy will be greatest with VF (kinematic) visual biofeedback. Patients Nineteen patients with dysphagia post stroke. Setting Outpatient swallowing research laboratory. Design Randomized clinical trial. Methods Patients underwent 2 study phases. Phase 1: first demonstrated ability to perform the vLVC accurately. Phase 2: vLVC training. Participants were randomized into three biofeedback groups including the ssEMG group (ssEMG biofeedback in both phases), the VF group (VF biofeedback in both phases), and the mixed group (VF phase 1, ssEMG phase 2). To promote the best vLVC performance, a clinician provided real‐time, verbal cueing using only the visual biofeedback type also seen by the patient, although both VF and ssEMG were recorded for all participants. Main Outcome Measure Patient performance accuracy and clinician feedback accuracy for performing the vLVC maneuver. Results Both accuracy of vLVC training performance and clinician feedback accuracy were worse in the ssEMG group compared with the VF and mixed groups (P < .001). Conclusions Swallowing airway protection requires precisely timed movements of small, hidden laryngeal and pharyngeal structures. Kinematic biofeedback (VF) may be required, at some point, to ensure that target swallowing movements are being trained during rehabilitation, rather than maladaptive movements. Level of Evidence I.