Mission-Critical Push-to-Talk (MCPTT) Statistics (2026): 55+ Data Points on Latency KPIs, FirstNet and Global MCX Networks, and Interoperability Testing

Mission-critical push-to-talk statistics 2026: NIST measured 103.9 ms LTE MCPTT access delay vs a 300 ms 3GPP target, plus FirstNet, UK ESN and MCX test data.

In NIST’s first measurements of a standards-based LTE mission-critical push-to-talk (MCPTT) system, access delay came in at 103.9 ± 4.1 ms, about a third of the 300 ms ceiling that 3GPP sets for the time between pressing the talk button and being cleared to speak. That one lab number captures the promise of broadband push-to-talk. The field picture is more mixed: AT&T says FirstNet now supports nearly 8 million connections, the UK has budgeted 19.1 billion pounds for its Emergency Services Network through 2043/44, and the latest ETSI interoperability event passed 92% of its tests, not 100%. This roundup focuses on broadband voice performance and MCX standards (MCX is the 3GPP umbrella for mission-critical push-to-talk, data and video); for radio system counts and the P25 upgrade cycle, see our land mobile radio and P25 migration statistics. We aggregated data from 3GPP, NIST’s Public Safety Communications Research division (PSCR), the FirstNet Authority, AT&T congressional testimony, the UK Home Office and National Audit Office, ETSI, Erillisverkot, an APEC submission by Korea, and peer-reviewed testbed studies.

TL;DR

  • MCPTT access time must be under 300 ms for 95% of requests (99% on emergency calls), and mouth-to-ear latency under 300 ms for 95% of voice bursts (3GPP, TS 22.179).
  • LTE MCPTT access delay measured 103.9 ± 4.1 ms, while P25 trunked Phase 2 radio showed 601.9 ms mouth-to-ear latency (NIST PSCR, NIST IR 8328 2020).
  • FirstNet grew from about 7 million connections in March 2025 to nearly 8 million in February 2026 (AT&T, House Energy and Commerce testimony 2026).
  • Over 2,600 temporary coverage solutions were triaged and deployed in fiscal 2025 (FirstNet Authority, Annual Report FY2025).
  • UK ESN whole life budget: 19.1 billion pounds to 2043/44, 6.1 billion pounds spent to date (Home Office, ESMCP 2025 Programme Business Case Summary).
  • South Korea’s PS-LTE Safe-Net cost approximately 1.1 billion US dollars and launched in May 2021 (Korea, APEC SDMOF submission 2025).
  • Virve 2 broadband subscribers equal one fifth of Finland’s legacy Virve base (Erillisverkot, Strategy and Business Review 2025).
  • France targets 300,000 users on its new Réseau Radio du Futur by 2027 (Acteurs Publics, reporting the Ministry of the Interior 2025).
  • ETSI’s 10th MCX Plugtests hit a 92% interoperability success rate across 356 tests (ETSI, 10th MCX Plugtests announcement 2026).

1. The 3GPP Latency Targets: What “Mission Critical” Means in Milliseconds

Mission-critical push-to-talk is defined by its timing rules. 3GPP TS 22.179 sets four key performance indicators (KPIs): access time (from pressing the push-to-talk button until the user gets the signal to speak), end-to-end access time, mouth-to-ear latency (from the moment a speaker talks until the listener’s speaker plays it back) and late call entry time (how long a user joining a call already in progress waits before hearing it). The headline target is an access time below 300 ms for 95% of requests, tightened to 99% for emergency group calls and imminent peril calls (calls placed when a responder faces immediate danger).

The fine print matters as much as the number. The KPIs are to be measured with negligible backhaul delay and under traffic load not exceeding 70% of each network node’s capacity, which describes a healthy network, not one saturated during a disaster. The standard has also kept moving: Release 13 was the specification’s initial planned release, and version 21.0.0 was uploaded on October 2, 2026 for Release 21, per the 3GPP specification page for TS 22.179.

MetricValueSource
MCPTT access time (KPI 1): group calls with no acknowledgement requested vs emergency group calls and imminent peril callsLess than 300 ms for 95% of all requests vs less than 300 ms for 99% of all requests3GPP, TS 22.179 (MCPTT Stage 1)
End-to-end MCPTT access time (KPI 2), users under coverage of the same networkLess than 1000 ms3GPP, TS 22.179 (MCPTT Stage 1)
Mouth-to-ear latency (KPI 3)Less than 300 ms for 95% of all voice bursts3GPP, TS 22.179 (MCPTT Stage 1)
Maximum late call entry time (KPI 4), without vs with application layer encryptionLess than 150 ms vs less than 350 ms3GPP, TS 22.179 (MCPTT Stage 1)
Extra performance allowance when a transcoder (a voice format converter) sits in the bearer path; initial and trailing lost audio permittedNo more than 40 ms; 0 ms3GPP, TS 22.179 (MCPTT Stage 1)
Latest version of TS 22.17921.0.0, uploaded October 2, 2026 (Release 21)3GPP, Specification Portal 2026

Context note: the 300 ms access time and 300 ms mouth-to-ear targets are separate budgets. A system can meet one and miss the other, which is why NIST measures them independently (see section 2).

2. Measured Performance: Lab and Testbed Results Against the 300 ms Bar

The best public measurement still comes from NIST PSCR, and it is encouraging. In initial testing of an LTE MCPTT system, access delay was 103.9 ± 4.1 ms with a 95% level of confidence, at NIST’s 85% raw intelligibility threshold (the point at which the test speech is judged understandable). In the same report, P25 trunked Phase 2 radio showed 601.9 ms of mouth-to-ear latency unencrypted, twice the 3GPP broadband target (601.9 / 300 = 2.0, our arithmetic). In other words, the broadband standard asks for faster voice delivery than trunked digital radio achieved in NIST’s own lab series.

Independent testbeds are more cautious. A 2019 IEEE Access study modeling MCPTT over LTE concluded that a current LTE network “is not yet able to guarantee the compliance with the KPIs defined by 3GPP under any circumstances”, and in one of its scenarios only access time (KPI 1) missed the 300 ms bar. A 2025 Fraunhofer FOKUS and TU Berlin proof of concept running open-source MCPTT on a private 5G network reported initial results that did not fully satisfy the defined limits. MCPTT rides on the same LTE and 5G voice architecture that carries consumer calls, a rollout tracked in our VoLTE and VoNR deployment statistics. The NIST numbers are in NIST IR 8328, and the LTE modeling study is the IEEE Access paper by Sanchoyerto et al..

MetricValueSource
LTE MCPTT access delay, initial test at 85% raw intelligibility (95% confidence interval)103.9 ± 4.1 msNIST PSCR, NIST IR 8328 2020
P25 direct mode (radio to radio, no network), unencrypted: access delay vs mouth-to-ear latency126.9 ± 8.8 ms vs 243.3 ± 0.3 msNIST PSCR, NIST IR 8328 2020
P25 trunked Phase 2, unencrypted: access delay vs mouth-to-ear latency655.1 ± 6.3 ms vs 601.9 ± 16.1 msNIST PSCR, NIST IR 8328 2020
P25 trunked Phase 2, encrypted: end-to-end access time1337.3 ± 28.0 msNIST PSCR, NIST IR 8328 2020
Access delay added by switching on encryption in P25 systems30-50 msNIST PSCR, NIST IR 8328 2020
LTE packet delay budget for MCPTT traffic, control plane vs user plane (voice)60 ms vs 75 msSanchoyerto et al., IEEE Access 2019
Expected latency reduction from hosting MCPTT at the network edge (MEC), modeled, each wayAt least 40%Sanchoyerto et al., IEEE Access 2019
Minimum group size used as the 3GPP scalability benchmark in a 5G capacity study2000 MCPTT users within an MCPTT groupSantiago et al., IEEE Access 2024

Context note: NIST’s LTE figure comes from its initial LTE MCPTT testing in a lab setup, so it is a proof that the target is reachable, not a field average. Most recent available data: NIST PSCR, 2020. The measurement method itself dates to NIST IR 8206 (2018), and capacity limits per cell are analyzed in the IEEE Access 5G capacity study.

3. FirstNet Scale and Coverage: Connections, Footprint and Indoor Reach

FirstNet is the U.S. nationwide public safety broadband network, built by AT&T under a 25-year contract with the FirstNet Authority on 20 MHz of dedicated Band 14 spectrum (a 700 MHz block reserved for public safety). Connections rose from about 7 million in March 2025 to nearly 8 million by February 2026, a gain of roughly 1 million in about 11 months (our arithmetic from two rounded figures). Growth is no longer only about subscribers: the money is shifting toward coverage depth, indoors and in rural areas.

Coverage is the selling point for push-to-talk users, because a talkgroup is only as good as its weakest signal. The FirstNet Authority Annual Report FY2025 puts the footprint above 2.99 million square miles; against the 3.8 million square miles of total U.S. area cited by RCR Wireless, that is about 79% (2.99 / 3.8, our arithmetic). The detail on tribal lands and new sites comes from AT&T’s February 2026 House testimony. On the dispatch side of the same voice chain, see our 911 text and voice dispatch statistics.

MetricValueSource
FirstNet connections, March 2025 vs February 2026About 7 million vs nearly 8 millionRCR Wireless reporting AT&T data 2025; AT&T, House Energy and Commerce testimony 2026
FirstNet network coverage, fiscal year 2025More than 2.99 million square milesFirstNet Authority, Annual Report FY2025
Terrestrial wireless coverage added in the year to March 202520,000 square milesRCR Wireless reporting AT&T data 2025
Increase in FirstNet coverage on tribal lands across the United StatesMore than 64%AT&T, House Energy and Commerce testimony 2026 (vendor-reported)
FirstNet tower sites launched in rural and tribal areas in the initial build vs additional Band 14 sites committed over 2 years2,000 vs 1,000AT&T, House Energy and Commerce testimony 2026 (vendor-reported)
Indoor coverage units delivered to primary public safety agencies, and indoor area covered5,546 units, up to approximately 85 million square feetFirstNet Authority, Annual Report FY2025

Context note: a connection is a device line, not a person, so the count is not a measure of responders using push-to-talk. FirstNet’s first standards-based push-to-talk service was announced in March 2020 as the first nationwide mission-critical standards-based PTT solution in the U.S., according to the FirstNet Authority; we found no public count of MCPTT users since.

4. FirstNet Under Load: Disasters, Priority Access and the Money Behind It

Latency targets assume a calm network; disasters are where they get tested. The FirstNet Authority FY2025 report counts over 2,600 temporary coverage solutions triaged and deployed in fiscal 2025, backed by a fleet of over 190 deployables (portable cell sites brought in when fixed towers are down or overloaded). Hurricane Helene alone produced 217 requests for emergency support. When towers fail, broadband push-to-talk is only as available as these stopgaps make it, so the deployable fleet matters as much as the 300 ms targets in section 1.

Priority access is the other half of the story. During one event in Butler, Pennsylvania, AT&T says more than 1,200 instances of First Priority (FirstNet’s feature that moves responders ahead of commercial traffic) safeguarded critical calls, and over 11,000 voice and data sessions were completed in a single hour. Long-term funding hinges on Congress: the FirstNet Authority sunsets on February 22, 2027 without action, and H.R. 7386, which passed the House in April 2026, would extend it through September 2037. Warnings to the public run on a separate track, covered in our emergency alert accessibility statistics.

MetricValueSource
FirstNet deployable fleet vs temporary coverage solutions triaged and deployed, fiscal year 2025Over 190 vs over 2,600FirstNet Authority, Annual Report FY2025
Hurricane Helene: requests for emergency support vs agencies served by dedicated deployables217 vs 66FirstNet Authority, Annual Report FY2025
Texas Hill Country floods, July 2025: support requests vs assets deployed24 vs 36FirstNet Authority, Annual Report FY2025
Butler, Pennsylvania event: First Priority instances vs voice and data sessions in one hourMore than 1,200 vs over 11,000AT&T, House Energy and Commerce testimony 2026 (vendor-reported)
Statutory FirstNet allocation from auction proceeds vs AT&T capital invested to date vs AT&T capital expected over the contract$6.5 billion vs more than $14 billion vs approximately $40 billionAT&T, House Energy and Commerce testimony 2026
FirstNet Authority cash balance on September 30, 2025, and AT&T payment in fiscal 2025$682 million and $420 millionFirstNet Authority, Annual Report FY2025

Context note: AT&T’s testimony puts Hurricanes Helene and Milton together at more than 260 emergency requests, which matches the Authority’s separate counts of 217 (Helene) and 48 (Milton), a total of 265 (our arithmetic).

5. UK Emergency Services Network: Budget, Options and the Airwave Overhang

The UK is replacing its Airwave TETRA radio network with a 4G mission-critical service, the Emergency Services Network (ESN). The Home Office’s 2025 Programme Business Case summary, issued in July 2026, sets a whole life budget of 19.1 billion pounds to 2043/44, with 6.1 billion pounds already spent. Its preferred option, continuing ESN, costs slightly less in present value than stopping ESN and keeping Airwave to 2043/44, and returns far higher benefits.

The comparison is less about building a new network than about the cost of keeping the old one alive. Legacy Airwave costs fall from about 8.1 billion pounds under the stop option to about 1.9 billion pounds if ESN continues. The National Audit Office warned in 2023 that keeping Airwave into the 2030s would cost at least 250 million pounds a year, and that planning permission was still needed for 42 of 292 remote area sites (masts built to extend coverage into rural Britain). Neither document gives a firm Airwave switch-off date; the business case says only that it will close “when it is safe to do so”. See the NAO progress report press release.

MetricValueSource
ESN whole life budget to 2043/44 vs spent to date19.1 billion pounds vs 6.1 billion poundsHome Office, ESMCP 2025 Programme Business Case Summary (2026)
Total cost, present value: stop ESN and keep Airwave to 2043/44 vs continue ESN9.47 billion pounds vs 9.35 billion poundsHome Office, ESMCP 2025 Programme Business Case Summary (2026)
Legacy Airwave costs: stop ESN option vs continue ESN optionAbout 8.1 billion pounds vs about 1.9 billion poundsHome Office, ESMCP 2025 Programme Business Case Summary (2026)
Total benefits, present value: stop ESN vs continue ESN (net present social value of continuing, relative to baseline)1.73 billion pounds vs 4.62 billion pounds (3.00 billion pounds)Home Office, ESMCP 2025 Programme Business Case Summary (2026)
Optimism bias allowance (extra cost added for the usual underestimate): stop ESN vs continue ESNAbout 8% vs about 12% of total costHome Office, ESMCP 2025 Programme Business Case Summary (2026)
Spent maintaining Airwave by March 2023 vs cost of keeping it into the 2030s2.9 billion pounds vs at least 250 million pounds a yearNational Audit Office, Progress with delivering the ESN 2023
Remote area sites still needing planning permission, 202342 of 292National Audit Office, Progress with delivering the ESN 2023
Airwave turn-off date: original plan, first extension, second extension2020, 2022, 2026 (later still expected)National Audit Office, Progress with delivering the ESN 2023

Context note: the business case extends the Mobile Services contract with EE (BT) through to 2031, and the Competition and Markets Authority charge control on Airwave pricing ends in 2029. The NAO figures date from March 2023 and are the most recent available audit data.

6. South Korea, Finland and France: How Other MCX Networks Compare

Outside the U.S. and UK, three national builds show three models. South Korea’s Safe-Net, a nationwide PS-LTE (public safety LTE) network, launched in May 2021 at a total cost of approximately 1.1 billion US dollars, on dedicated 718-728 MHz and 773-783 MHz bands, with three operation centers in Seoul, Daegu and Jeju. Finland took a different route: Virve 2 is built on a commercial operator’s 4G and 5G radio access network, and its service area covers approximately 95% of the country’s land area.

Adoption is the slow part everywhere. Finland’s legacy TETRA-based Virve still grew to almost 60,000 subscriptions in 2025, while Virve 2 broadband subscribers number one fifth of that base, and about half of them use the group voice service. France’s Réseau Radio du Futur (RRF) finished a test phase with 800 agents in late 2025 and targets 300,000 users by 2027. Sources: the APEC submission by Korea, the Erillisverkot Annual Report 2025, the Erillisverkot Strategy and Business Review 2025, Erillisverkot on Virve 2 coverage and Acteurs Publics on the RRF.

MetricValueSource
South Korea Safe-Net total costApproximately 1.1 billion US dollarsKorea, APEC SDMOF submission 2025
South Korea Safe-Net frequency bands718-728 MHz and 773-783 MHzKorea, APEC SDMOF submission 2025
South Korea Safe-Net terminals in use at the May 2021 opening vs planned by end of 2021About 90,000 vs more than 150,000Asia Economy, reporting the Ministry of the Interior and Safety opening ceremony 2021
Finland Virve (TETRA) subscriptions, end of 2025 vs 2024Almost 60,000 vs almost 55,000Erillisverkot, Annual Report 2025
Finland Virve 2 broadband subscribers as a share of legacy Virve subscribers, and share using the group voice serviceOne fifth; about halfErillisverkot, Strategy and Business Review 2025
Finland Virve 2 service area, July 2025Approximately 95% of Finland’s land areaErillisverkot, Virve 2 ready for deployment 2025
France RRF: agents in the test phase (June to November 19, 2025) vs users targeted by 2027800 vs 300,000Acteurs Publics, reporting Acmoss and the Ministry of the Interior 2025

Context note: the Korean terminal counts date from May 2021 and are the most recent available figures we could verify; we found no official 2025 count of Safe-Net terminals or user agencies. Erillisverkot also granted a permanent Virve licence to 21 new organisations in 2025, and its revenue rose to 129 million euros (Erillisverkot, Annual Report 2025).

7. Interoperability Testing: ETSI MCX Plugtests Results

Standards only help if products from different vendors work together, and ETSI’s MCX Plugtests (week-long multi-vendor interoperability test events) are the public scoreboard. The 10th event, held May 18-22, 2026 in Sophia Antipolis, reached a 92% interoperability success rate across 356 tests in 112 test sessions, with more than 240 participants. The 9th event, hosted by Texas A&M in February 2025, scored 83.3%: 145 of 174 test executions passed.

The trend line looks like a decline from the 97.6% of 2021, but the events are not like for like. Earlier editions counted executed use cases in the thousands, while recent ones report a few hundred test executions on newer 3GPP releases (Release 18 in 2025, Releases 18 and 19 in 2026, per the 3GPP MCX page). The test specification itself keeps growing, from 402 test cases in 2025 to 466 in 2026, which means more features to break. Reports: the 9th MCX Plugtests Report, ETSI’s 10th Plugtests announcement, and the 8th, 7th and 6th event releases.

MetricValueSource
10th MCX Plugtests, May 2026: success rate, tests, test sessions, participants92%; 356 tests; 112 sessions; more than 240 participantsETSI, 10th MCX Plugtests Report announcement 2026
9th MCX Plugtests, February 2025: success rate (passed of executed), test sessions, people involved83.3% (145 of 174); 146 sessions; 221 peopleETSI, 9th MCX Plugtests Report 2025
9th MCX Plugtests: application server vendors vs client vendors vs remote labs connected17 vs 20 vs 29ETSI, 9th MCX Plugtests Report 2025
8th MCX Plugtests, November 2023 (3GPP Release 17): success rate, use cases executed, delegates and vendors95%; over 1,500; more than 180ETSI, 8th MCX Plugtests announcement 2023
7th MCX Plugtests, December 2022 (3GPP Release 17): success rate, use cases executed, delegates and vendors96.0%; over 1,200; more than 150ETSI, 7th MCX Plugtests announcement 2022
6th MCX Plugtests, December 2021 (3GPP Release 16): success rate, test cases, test sessions, participants97.6%; around 1,300; 205; 150ETSI, 6th MCX Plugtests announcement 2021
Test cases in the ETSI MCX test specification, 2025 vs 2026402 vs 466ETSI, 9th MCX Plugtests Report 2025; ETSI, 10th MCX Plugtests announcement 2026

Context note: the 6th and 7th event figures are older than three years. Most recent available data for those editions: ETSI, 2021 and 2022.

Summary: Mission-Critical Push-to-Talk by the Numbers

MetricValueSource
MCPTT access time target (KPI 1): standard calls vs emergency and imminent peril callsUnder 300 ms for 95% vs 99% of requests3GPP, TS 22.179
Mouth-to-ear latency target (KPI 3)Under 300 ms for 95% of voice bursts3GPP, TS 22.179
LTE MCPTT access delay, initial NIST test103.9 ± 4.1 msNIST PSCR, NIST IR 8328 2020
P25 trunked Phase 2 mouth-to-ear latency, unencrypted601.9 ± 16.1 msNIST PSCR, NIST IR 8328 2020
Minimum MCPTT group size used as a scalability benchmark2000 usersSantiago et al., IEEE Access 2024
FirstNet connections, February 2026Nearly 8 millionAT&T, House Energy and Commerce testimony 2026
Temporary coverage solutions deployed, fiscal year 2025Over 2,600FirstNet Authority, Annual Report FY2025
AT&T capital expected over the FirstNet contractApproximately $40 billionAT&T, House Energy and Commerce testimony 2026
UK ESN whole life budget to 2043/4419.1 billion poundsHome Office, ESMCP 2025 Programme Business Case Summary (2026)
UK ESN spent to date6.1 billion poundsHome Office, ESMCP 2025 Programme Business Case Summary (2026)
South Korea Safe-Net total costApproximately 1.1 billion US dollarsKorea, APEC SDMOF submission 2025
Finland Virve 2 service areaApproximately 95% of land areaErillisverkot, Virve 2 ready for deployment 2025
Finland Virve subscriptions, end of 2025Almost 60,000Erillisverkot, Annual Report 2025
France RRF users targeted by 2027300,000Acteurs Publics, reporting the Ministry of the Interior 2025
ETSI 10th MCX Plugtests success rate92%ETSI, 10th MCX Plugtests announcement 2026
ETSI 9th MCX Plugtests success rate83.3%ETSI, 9th MCX Plugtests Report 2025

Methodology and Sources

Last updated: October 24, 2026. We update this roundup quarterly, and the next refresh is expected when AT&T reports its fourth-quarter 2026 results and ETSI publishes the report from its next MCX Plugtests.

Try VoxBooster — 3-day free trial.

Real-time voice cloning, soundboard, and effects — wherever you already talk.

  • No credit card
  • ~30ms latency
  • Discord · Teams · OBS
Try free for 3 days