Background: Vascular Ehlers-Danlos syndrome is a rare inherited connective tissue disease secondary to mutations within the COL3A1 gene. The diagnosis of vascular Ehlers-Danlos syndrome is challenging, and patient selection for genetic testing relies on diagnostic criteria, which have never been evaluated. Methods: All patients seen at a dedicated tertiary referral center for a suspicion of vascular Ehlers-Danlos syndrome between January 2001 and March 2016 were retrospectively included in a diagnostic accuracy study. Major and minor diagnostic criteria of the Villefranche classification were tested for sensitivity, specificity, positive and negative predictive values, according to results of genetic testing. Results: N=519 patients were eligible for analysis dividing into n=384 probands and n=135 relatives. A pathogenic COL3A1 variant was identified in n=165 (31.8%) patients. The Villefranche criteria were met for n=248 patients with a sensitivity of 79% (95% CI, 0.72-0.85) and a negative predictive value of 87% (95% CI, 0.83-0.91). Diagnostic accuracy was highest for symptomatic probands (sensitivity 92%; negative predictive value 95%) with limited specificity (60%). Probands ≤25 years had the worst diagnostic performance. The revised diagnostic Criteria (2017) were less accurate than the Villefranche classification (overall diagnostic odds-ratio, 4.17 versus 7.8; probands diagnostic odds-ratio, 4.04 versus 18.1; and probands ≤25 years diagnostic odds-ratio, 2.36 versus 5.1) mainly due to a lack of sensitivity. Conclusions: The Villefranche criteria provide accurate detection of symptomatic probands in specialized practice but have limited specificity. The revised diagnostic criteria for vascular Ehlers-Danlos syndrome have increased specificity, but its overall performance is poorer. The early clinical diagnosis of probands without family history is not addressed by both diagnostic classifications.