Can I dial 911 from a VoIP phone?

911 on a VoIP line is a regulated obligation rather than a product feature. The rules state that providers of interconnected VoIP service must, as a condition of providing service to a consumer, provide that consumer with E911 service, and must transmit every 911 call together with automatic number identification and location information to the public safety answering point serving the caller's location [1]. For essentially all residential and business VoIP sold in the United States the call connects, so the more useful question is what the dispatcher sees when it arrives.

Getting the call answered

Delivering the audio is routine. The provider hands the call to the emergency network using the calling number, and a pseudo-number where one is needed, so that it lands at the answering point serving the caller's dispatchable location [1]. In a January 2024 order the FCC cited NENA data indicating that 5,748 PSAPs operate in the United States [10], which is why routing has to be driven by a location record instead of a single national queue. Where the provider cannot obtain or confirm a location, the rules permit the call to go first to a national emergency call center whose staff collect the address by voice [1].

Getting the address right

A copper line reported an address because the wire pair physically ended there. A VoIP number carries no geography of its own, so the rules substitute a Registered Location, defined as the most recent information a provider has that identifies the physical location of an end user [2]. Almost everything that goes wrong with a VoIP 911 call traces back to that record being out of date or missing the floor and suite.

Outbound-only services are covered too

The usual definition of interconnected VoIP covers services that both place and receive calls on the public phone network. For emergency calling the FCC widened it: solely for purposes of compliance with its 911 obligations, an interconnected VoIP service includes a service that permits users generally to terminate calls to the public switched telephone network, whether or not it can receive them [2]. A dialer built into a business application therefore carries the same 911 duties, even though a call taker cannot ring it back.

Yes, and on a fixed line the provider carries the entire location burden, because the rules require providers of fixed interconnected VoIP services to supply automated dispatchable location with each 911 call [1]. Dispatchable location means the validated street address of the calling party plus whatever suite, apartment, or similar detail is needed to identify where that party is [2], so a complete record names the room and not only the building.

This is the closest VoIP gets to the behavior of an old copper line, although the mechanism underneath is different. The address lives in a database record the provider populates and the 911 center queries, while the call itself is routed on the calling number through the dedicated emergency network [1]. One limit is written into the rule rather than into any carrier's product: if a 911 center is not capable of receiving and processing the calling number or the location, the provider is not required to deliver them [1].

Because the whole arrangement rests on a stored record, it is worth checking that record directly. Most carriers answer a call to 933 with an automated readback of the calling number and the emergency address registered to it, and that call never reaches a dispatcher. Telnyx tells customers to use 933 to confirm the address read back matches what they registered, and warns against placing live 911 calls for routine testing [9]. Running that check at installation and again after any move catches the two problems that cause most trouble: an address nobody updated, and a suite number that fell off between the order form and the location database.

Yes, though how well it works depends on how your provider tracks movement. For non-fixed interconnected VoIP, meaning service capable of being used from more than one location, the rules ask first for automated dispatchable location where that is technically feasible. Where it is not, the provider may instead supply a Registered Location, or alternative location information, or route the caller to a national emergency call center [1].

The Registered Location path carries three conditions that are easy to skim past. The provider has to obtain the address from the customer before service starts, has to offer at least one way to update it that needs only the equipment already required to use the service, and has to identify when a 911 call is placed from somewhere other than the registered address, then either prompt the customer for a new one or update it without asking [1]. A laptop signed in from a hotel room is precisely the situation those conditions were written for.

Providers satisfy the detection requirement with network evidence rather than satellite positioning. RingCentral has administrators register wireless access points, IP address ranges, and Ethernet switch details, and the app or device location is then sent to the appropriate emergency service agency when a call is placed from it [8]. Vendor descriptions of nomadic E911 build on the same idea, carrying a current location object inside the SIP signaling so the call can route on where the person is at that moment [13][1]. If you work from more than one place, the address field needs periodic attention, because it only ever reflects whatever was entered last.

Two federal laws add obligations that consumer VoIP service does not carry. Kari's Law requires that a user be able to initiate a call to 911 directly from any station equipped with dialing facilities, without an additional digit, code, or prefix such as the trunk-access digit 9, even where other outbound calls still need one [3]. It applies to a multi-line telephone system manufactured, imported, offered for first sale or lease, first sold or leased, or installed after February 16, 2020 [5]. The same rule requires a notification to a central point at the facility, or to another person or organization anywhere, sent at the same time as the call, without delaying it, and to somewhere a person is likely to see or hear it [3].

What counts as a multi-line telephone system

The definition is deliberately wide, covering network-based and premises-based systems including Centrex and VoIP along with PBX, hybrid, and key telephone systems, whether the operator is a business, a government agency, or a nonprofit [2]. The FCC read it to include cloud-based technology and over-the-top applications, and to include enterprise systems that allow outbound calls to 911 without giving the 911 center a way to call back [5].

The dispatchable location deadlines by device type

Section 506 of RAY BAUM'S Act produced a schedule keyed to where the device sits. On-premises fixed telephones had to provide automated dispatchable location by January 6, 2021, while on-premises non-fixed devices had until January 6, 2022 with a fallback to manual end-user updates or alternative location information. Off-premises devices share that 2022 date, and where automatic location is not technically feasible they may rely on manual updates or on enhanced location information, which the rule allows to be coordinate-based [3].

What compliance looks like in an admin console

Microsoft Teams is a useful illustration because its model maps onto the rule almost directly. Teams keeps an emergency address, meaning the civic street address, separate from a place, meaning a floor or office number, and joins the two into an emergency location with its own identifier that gets assigned to a user or a site [6]. An address has to be marked validated before it can be assigned, and dynamic emergency calling requires a geo code on that address, which is why Microsoft advises creating addresses through the map search in the Teams admin center [6]. Dynamic emergency calling is available across Microsoft Calling Plans, Operator Connect, Teams Phone Mobile, and Direct Routing [6].

Apps that connect users only to each other, without ever touching the public phone network, fall outside the interconnected VoIP definition and generally offer no path to 911 [2]. The obligation attaches to services that let users place calls to ordinary telephone numbers, so a messaging app whose voice feature stops at its own user base is not covered by these rules at all.

Most services sit between those extremes, and that is where people get caught out. Google Voice supports calls to emergency services, but only on Google Workspace add-on accounts managed by a work or school and on standalone Voice Starter or Standard accounts, and Google states that in a power outage, a loss of internet connection, or a Google Voice service outage, emergency calling will not be available [7]. Google also publishes a printable warning label to keep near the phone [7].

Texting is a separate gap. The FCC's text-to-911 requirements apply to a covered text provider, defined as all CMRS providers together with providers of interconnected text messaging services capable of exchanging messages with substantially all text-capable United States numbers [4]. A VoIP desk phone is neither, so texting 911 from one is usually impossible even in a county where it works from a cell phone.

Redundancy handles this cheaply. Any phone can reach emergency services, including a mobile handset with no account or service plan behind it [7], so a spare handset covers a site whose voice service depends on broadband staying up.

The failure modes are documented well enough that the FCC made providers warn customers about them by name. Every interconnected VoIP provider has to advise every subscriber, prominently and in plain language, of the circumstances under which E911 may be unavailable or limited, and the rule lists relocation of the customer's equipment, use of a non-native telephone number, broadband connection failure, loss of electrical power, and delays in making a dispatchable location available through the ALI database [1]. Providers also have to record each subscriber's acknowledgment of that advisory and either distribute warning stickers or give conspicuous notice another way [1].

A stale address sends responders to the wrong door

Nothing in the system independently confirms that the address on file is where the phone is. A provider of non-fixed service does have to identify a 911 call placed from somewhere other than the registered address and respond to it [1], but that detection depends on network data the provider can actually observe, so a device on an unfamiliar network can still arrive carrying an old address.

The 911 center may not be able to use what you send

Where an answering point cannot receive and process the calling number or the location, the provider is excused from delivering them [1]. Capability varies by jurisdiction and is not something a customer can configure around, which is one reason it remains worth stating your address out loud on any emergency call.

Outages reach the emergency network itself

The 911 network behind your provider is mid-rebuild, and the FCC treats outage risk in that rebuild as a live problem. Its Next Generation 911 rules, effective August 10, 2026, require entities essential to delivering emergency calls to adopt measures safeguarding network reliability and reducing the risk of 911 outages, and require certain providers to report on interoperability [11]. A companion notice adopted on June 25, 2026 seeks comment on interoperability testing and on direct video calling into NG911 networks [12]. None of this is under a customer's control, which is the argument for a second path at any single-connection site.

A 933 test call verifies the record in under a minute

Dialing 933 reads back the calling number and the registered emergency address without dispatching anyone [9], and platforms such as Teams let administrators define how a 933 call is handled by security desk notifications so the test does not create noise [6]. Scheduling it yearly per site, and after every office move, costs almost nothing.

It is not the same as wireless 911 location

The dispatchable location standard governs VoIP and multi-line telephone systems. Commercial Mobile Radio Service providers are carved out of that definition and instead convey the location required by the wireless rules [2], so a cell phone and a softphone sitting on the same table can present a caller's position in different ways.

It is not text-to-911

Text obligations sit with CMRS providers and with interconnected text messaging services [4]. A VoIP handset is outside that category, so text-to-911 is generally unavailable from one regardless of what the local 911 authority supports.

It is not a live location sensor

A Registered Location is a stored record, described in the rules as the most recent information the provider has that identifies the end user's physical location [2]. It reflects the last update rather than the present moment, which is exactly why the update and detection requirements exist alongside it.

It is not something a covered provider can leave out

E911 has to be provided as a condition of selling interconnected VoIP service [1], so no covered provider legitimately sells a cheaper plan with emergency calling stripped out. A service that cannot reach 911 is signaling that it falls outside the definition entirely.