Teredo clients, when source routing is enabled, recognize a Routing header in an encapsulated IPv6 packet and send the packet to the next hop, which might allow remote attackers to bypass policies of certain Internet gateways that drop all source-routed packets.

Teredo creates trusted peer entries for arbitrary incoming source Teredo addresses, even if the low 32 bits represent an intranet address, which might allow remote attackers to send IPv4 traffic to intranet hosts that use non-RFC1918 addresses, bypassing IPv4 ingress filtering.

Teredo clients, when located behind a restricted NAT, allow remote attackers to establish an inbound connection without the guessing required to find a port mapping for a traditional restricted NAT client, by (1) using the client port number contained in the Teredo address or (2) following the bubble-to-open procedure.

Teredo clients, when following item 6 of RFC4380 section 5.2.3, start direct IPv6 connectivity tests (aka ping tests) in response to packets from non-Teredo source addresses, which might allow remote attackers to induce Teredo clients to send packets to third parties.