Background: Temporal artery biopsy is considered the diagnostic gold standard for giant cell arteritis, despite approximately 39% of patients who are negative for the condition by biopsy subsequently being given a clinical diagnosis of giant cell arteritis. We aimed to assess the diagnostic accuracy of ultrasound examination in patients with suspected giant cell arteritis. Methods: In this prospective, multicentre, non-interventional, cohort study (evaluation of ultrasound's role in patients suspected of having extracranial and cranial giant cell arteritis; EUREKA), we consecutively recruited patients aged 50 years or older, with clinically suspected giant cell arteritis from three Danish hospitals (South West Jutland Hospital in Esbjerg, Silkeborg Regional Hospital, and Rigshospitalet, Glostrup). Participants had a bilateral ultrasound of the temporal, facial, common carotid, and axillary arteries. Ultrasounds were done by ultrasonographers who were systematically trained in vascular ultrasound using appropriate equipment and settings. Participants then had a temporal artery biopsy within 7 days of initiation of corticosteroid treatment. A blinded ultrasound expert assessed all ultrasound images. Ultrasound vasculitis was defined in cranial arteries as a homogeneous, hypoechoic, intimamedia complex thickness and a positive compression sign and as a homogeneous intimamedia complex of 1 mm in thickness or wider in the axillary arteries and of 1·5 mm thickness or wider in the common carotid artery. Participants were followed up at 6 months. During this 6 month period, clinicians were able to collect data from all clinical examinations to enable a full clinical diagnosis at 6 months. Clinical diagnosis was based on the expert opinion of the treating rheumatologist. The diagnostic criterion standard was diagnosis confirmed after 6 months of follow-up. We used logistic regression analyses to calculate the odds ratio and 95% CI of ultrasound as a predictor for giant cell arteritis. Findings: Between April 1, 2014, and July 31, 2017, 118 patients were screened for inclusion, of whom 106 had both ultrasound examinations and an eligible temporal artery biopsy and were included in the intention-to-diagnose population. The mean age was 72·7 years (SD 7·9), 63 (59%) participants were women, and 43 (41%) were men. Temporal artery biopsy was positive in 46 (43%) of 106 patients, and 62 (58%) of 106 patients had a clinically confirmed diagnosis of giant cell arteritis at 6 months (temporal artery biopsy sensitivity 74% [95% CI 62–84], specificity 100% [95% CI 92–100]). Cranial artery ultrasound was positive in all patients who had a positive temporal artery biopsy, and seven (58%) of 12 patients who were positive by ultrasound and negative by temporal artery biopsy were confirmed to have large-vessel giant cell arteritis via other imaging methods. The sensitivity of ultrasound diagnosis of giant cell arteritis was 94% (84–98) and specificity was 84% (70–93). Logistic regression analysis confirmed that ultrasound was the strongest baseline predictor for a clinically confirmed diagnosis of giant cell arteritis at 6 months (crude odds ratio 76·6 [95% CI 21·0–280·0]; adjusted for sex and age 141·0 [27·0–743·0]). Interpretation: Vascular ultrasound might effectively replace temporal artery biopsy as a first-line diagnostic method in patients suspected of having giant cell arteritis when done by systematically trained ultrasonographers using appropriate equipment and settings. Funding: The Institute for Regional Research at Hospital of Southwest Jutland, Esbjerg, Denmark.