Semi-implantable transcutaneous bone conduction devices are treatment options for conductive and mixed hearing loss (CHL/MHL). For counseling of patients, realistic simulation of the functional result is desirable. This study compared speech recognition in noise with a semi-implantable transcutaneous bone conduction device to external stimulation with a bone conduction device fixed by a headband. Eight German-language adult patients were enrolled after a semi-implantable transcutaneous bone conduction device (Bonebridge, Med-El) was implanted and fitted. Patients received a bone conduction device for external stimulation (Baha BP110, Cochlear) fixed by a headband for comparison. The main outcome measure was speech recognition in noise (Oldenburg Sentence Test). Pure-tone audiometry was performed and subjective benefit was assessed using the Glasgow Benefit Inventory and Abbreviated Profile of Hearing Aid Benefit questionnaires. Unaided, patients showed a mean signal-to-noise ratio threshold of 4.6 ± 4.2 dB S/N for speech recognition. The aided results were −3.3 ± 7.2 dB S/N by external bone conduction stimulation and −1.2 ± 4.0 dB S/N by the semi-implantable bone conduction device. The difference between the two devices was not statistically significant, while the difference was significant between unaided and aided situation for both devices. Both questionnaires for subjective benefit favored the semi-implantable device over external stimulation. We conclude that it is possible to simulate the result of speech recognition in noise with a semi-implantable transcutaneous bone conduction device by external stimulation. This should be part of preoperative counseling of patients with CHL/MHL before implantation of a bone conduction device.