What is T.38 protocol in simple terms?

T.38 is an ITU-T standard that converts fax signals into IP-friendly data packets for real-time transmission over the internet. It decodes the analog T.30 fax tones at a gateway, repackages them as fax-specific UDP packets with built-in redundancy, and reassembles them at the other end — all transparently to the fax machines involved.

Why does fax fail on VoIP without T.38?

Standard fax (T.30) relies on precise analog tones and strict timing. VoIP networks introduce packet loss, jitter, and codec compression — any of which disrupts the fax handshake. Even 1% packet loss can cause failures with G.711 passthrough. T.38 was designed specifically to tolerate these conditions, handling up to 10%+ packet loss with redundancy.

What is UDPTL in T.38?

UDPTL (UDP Transport Layer) is the preferred transport for T.38 packets. It uses a compact 2-byte header (versus RTP's 12-byte header) and supports packet redundancy — each packet carries copies of previous packets so the receiver can reconstruct lost data. UDPTL is used instead of RTP because TCP's retransmission delays break fax timing requirements.

Do both sides need to support T.38 for it to work?

Yes. Both the sending and receiving gateways must support T.38 and agree to use it via SIP re-INVITE negotiation. If any network segment between them lacks T.38 support, the call falls back to audio passthrough, reintroducing all the reliability problems T.38 was meant to solve.

Is T.38 required for cloud fax services like mFax?

No. Cloud fax services like [mFax](https://mfax.to) maintain their own dedicated fax infrastructure and handle all protocol negotiation internally. You never configure T.38 — you just upload a document and send. This is why cloud fax consistently delivers higher reliability than self-managed T.38 on VoIP.

T.38 Protocol Explained: Everything You Need to Know

T.38 is the ITU-T standard that makes real-time fax transmission possible over IP networks. This deep-dive covers how T.38 works, UDPTL packet structure, redundancy settings, SIP re-INVITE negotiation, and when to use T.38 versus G.711 or cloud fax.

Frequently Asked Questions

What is T.38 protocol in simple terms?: T.38 is an ITU-T standard that converts fax signals into IP-friendly data packets for real-time transmission over the internet. It decodes the analog T.30 fax tones at a gateway, repackages them as fax-specific UDP packets with built-in redundancy, and reassembles them at the other end — all transparently to the fax machines involved.
Why does fax fail on VoIP without T.38?: Standard fax (T.30) relies on precise analog tones and strict timing. VoIP networks introduce packet loss, jitter, and codec compression — any of which disrupts the fax handshake. Even 1% packet loss can cause failures with G.711 passthrough. T.38 was designed specifically to tolerate these conditions, handling up to 10%+ packet loss with redundancy.
What is UDPTL in T.38?: UDPTL (UDP Transport Layer) is the preferred transport for T.38 packets. It uses a compact 2-byte header (versus RTP's 12-byte header) and supports packet redundancy — each packet carries copies of previous packets so the receiver can reconstruct lost data. UDPTL is used instead of RTP because TCP's retransmission delays break fax timing requirements.
Do both sides need to support T.38 for it to work?: Yes. Both the sending and receiving gateways must support T.38 and agree to use it via SIP re-INVITE negotiation. If any network segment between them lacks T.38 support, the call falls back to audio passthrough, reintroducing all the reliability problems T.38 was meant to solve.
Is T.38 required for cloud fax services like mFax?: No. Cloud fax services like [mFax](https://mfax.to) maintain their own dedicated fax infrastructure and handle all protocol negotiation internally. You never configure T.38 — you just upload a document and send. This is why cloud fax consistently delivers higher reliability than self-managed T.38 on VoIP.