Paravertebral arteriovenous shunts (PVAVSs) are rare. Whether the intradural venous system is involved in drainage may lead to differences in clinical characteristics through specific pathophysiological mechanisms. This study aims to comprehensively evaluate the natural history and clinical outcomes of PVAVSs with or without intradural drainage.Sixty-four consecutive patients with PVAVSs from 2 institutes were retrospectively reviewed. Lesions were classified as type A (n=28) if the intradural veins were involved in drainage; otherwise, they were classified as type B (n=36). The clinical course from initial presentation to the last follow-up was analyzed.The patients with type A shunts were older at presentation (52.5 versus 35.5 years, P<0.0001) and more likely to have lower spinal segments affected than patients with type B PVAVSs (67.8% versus 13.9%, P=0.00006). After presentation, the deterioration rates related to gait and sphincter dysfunction were significantly higher in patients with type A than type B shunts (gait dysfunction: 71.8%/y versus 17.0%/y, P=0.0006; sphincter dysfunction: 63.7%/y versus 11.3%/y, P=0.0002). According to the angiogram at the end of the latest treatment, 79% of type A and 75% of type B PVAVSs were completely obliterated. If the lesions were partially obliterated, a significantly higher clinical deterioration rate was observed in patients with type A shunts than those with type B shunts (69.9%/y versus 3.2%/y, P=0.0253).Type A PVAVSs feature rapid progressive neurological deficits; therefore, early clinical intervention is necessary. For complex lesions that cannot be completely obliterated, surgical disconnection of all refluxed radicular veins is suggested.