Objectives To evaluate the efficacy and side effects of 'swinging eyelid' orbital decompression in patients with Graves' orbitopathy (GO). To calculate the incidence of postoperative new-onset diplopia (NOD) using a newly proposed scoring system for diplopia. Methods We reviewed the clinical data on proptosis, visual acuity, and diplopia in 104 consecutive patients (198 orbits) with GO, who underwent orbital decompression. A combined lateral canthal and inferior fornix incision ('swinging eyelid' approach) was used for removal of the medial wall, the orbital floor and, if indicated, the lateral wall. Indications for surgery were disfiguring/ congestive GO (DGO) in 79 patients (149 orbits) and compressive optic neuropathy (CON) in 25 patients (49 orbits). Diplopia was scored according to four grades. In both groups, the incidence of new-onset (continuous) diplopia (NOD), deterioration of diplopia (DOD), and improvement of diplopia (IOD) were calculated, using strictly defined criteria. Our data on NOD were compared to those from other series, after recalculation according to our criteria. Results The mean proptosis reduction was 4.6 mm (range 0-9.5 mm) after three-wall decompression (95 patients, 180 orbits) vs 3.1 mm (range 0-7 mm) after two-wall decompression (nine patients, 18 orbits). The visual acuity improved in 98% of the patients with CON. In patients with DGO, NOD occurred in 14%. In patients with CON, NOD was not observed, but DOD occurred in 41%. Our data compare favourably to the reported incidence of NOD after either transantral or transnasal decompression. Conclusions 'Swinging eyelid' orbital decompression is efficacious for proptosis reduction as well as for optic nerve decompression. A scoring system for standardized evaluation of diplopia is proposed.