Background Clinical observations and results from recent studies support the use of non-invasive vagus nerve stimulation (nVNS) for treating cluster headache (CH) attacks. This study compared nVNS with a sham device for acute treatment in patients with episodic or chronic CH (eCH, cCH). Methods After completing a 1-week run-in period, subjects were randomly assigned (1:1) to receive nVNS or sham therapy during a 2-week double-blind period. The primary efficacy endpoint was the proportion of all treated attacks that achieved pain-free status within 15 minutes after treatment initiation, without rescue treatment. Results The Full Analysis Set comprised 48 nVNS-treated (14 eCH, 34 cCH) and 44 sham-treated (13 eCH, 31 cCH) subjects. For the primary endpoint, nVNS (14%) and sham (12%) treatments were not significantly different for the total cohort. In the eCH subgroup, nVNS (48%) was superior to sham (6%; p < 0.01). No significant differences between nVNS (5%) and sham (13%) were seen in the cCH subgroup. Conclusions Combing both eCH and cCH patients, nVNS was no different to sham. For the treatment of CH attacks, nVNS was superior to sham therapy in eCH but not in cCH. These results confirm and extend previous findings regarding the efficacy, safety, and tolerability of nVNS for the acute treatment of eCH.