Skip to content

Glossary drawing · CPaaS & channels

Concatenated SMS

Also called: Long SMS · Multipart SMS · Segmented SMS · SMS concatenation

Note 1 —A long SMS split into segments of up to 153 GSM-7 or 67 UCS-2 characters, which a header lets the phone rejoin, each billed as a separate message.

Term
Concatenated SMS
Abbreviation
—
Category
CPaaS & channels
Level
Practitioner
Reviewed
Sheet no.
GLO-CPA-486
Drawn by
InfiQ Glossary

Drawing 1 · Interactive

Add one character and watch the SMS split

5 / 5

Encoding

GSM-7

7-bit alphabet · 160 per SMS, 153 per part

Characters

154

163 septets

SMS parts

2

153 per part

Left in part

143

before part 3

Billed as 2 SMS

Part 1153 / 153

Kesarbagh Sweets order 4821 | [Kaju katli x2] [Motichoor ladoo x1] | Pickup {Hazratganj} after 6 pm ~ show this SMS at the counter. Reply STOP t

Part 210 / 153

o opt out.

Header (UDH) — 7 septets of every part go to reassembly data.

How it splits

Part 1Kesarbagh Sweets order 4821 | [Kaju katli x2] [Motichoor ladoo x1] | Pickup {Hazratganj} after 6 pm ~ show this SMS at the counter. Reply STOP tPart 2o opt out.
  • Alternating parts
  • 2 septets
  • Forces UCS-2

Why

Past 160 septets the message is split into 2 parts of up to 153: each part gives 7 septets to a header that lets the phone stitch the parts back together. 9 extension characters (€ [ ] { } \ ^ ~ |) cost 2 septets each.

Pick a sample or type your own. The scene shows the encoding, the units used and how many billed segments the text becomes, and highlights the characters that force UCS-2.

Extension characters count double: GSM-7, 2 SMS parts.

What is concatenated SMS?

Concatenated SMS is the standard way to send a text longer than one SMS can carry. A single SMS has a payload of 140 octets: 160 characters in the GSM-7 alphabet, or 70 in UCS-2. Anything longer is cut into segments, each sent as its own SMS with a small User Data Header (UDH) in front of the text, and the receiving phone uses that header to put the pieces back in order and show one message.

The usual header is six octets: 05 (header length), 00 (information element: concatenation with an 8-bit reference), 03 (element length), then a reference number shared by every part, the total number of parts and this part's sequence number. Those six octets come out of the 140, which is why a GSM-7 segment holds 153 characters and a UCS-2 segment 67. A variant with a 16-bit reference (element 08) takes seven octets and leaves 152 or 66.

Concatenation happens after the encoding is decided: GSM-7 Encoding and UCS-2 Encoding set the alphabet, and concatenation does the arithmetic on top. One character outside GSM-7, such as a Hindi letter, an emoji or a curly quote, moves the whole message to UCS-2, so the same 150-character offer goes from one part to three. Operators and aggregators bill per segment and return a delivery report per segment, so a long text is really several SMS that happen to display as one.

Key takeaways

  • A concatenated SMS is one long text sent as several SMS segments that the phone stitches back into a single message.
  • Each segment gives 6 octets to a User Data Header, so parts hold 153 GSM-7 or 67 UCS-2 characters instead of 160 or 70.
  • Every segment is billed as a separate SMS: a 161-character English text costs two, and so does a 71-character Hindi one.
  • Over SMPP the UDH rides in short_message with esm_class 0x40, or SAR TLVs or message_payload let the SMSC split the text.
  • A lost part or a stripped header breaks reassembly, so keep business texts short and track delivery reports per segment.

Specifications

Concatenated SMS: key facts and limits

Single SMS payload
140 octets: 160 GSM-7 or 70 UCS-2 characters
Per segment
153 GSM-7 or 67 UCS-2 characters
UDH size
6 octets (8-bit reference) or 7 (16-bit)
UDH fields
Reference, total parts, sequence number
SMPP flag
esm_class 0x40 (UDHI) when the UDH is in short_message
Billing
Each segment billed as one SMS
Maximum parts
255 by the header; providers cap far lower

Operating sequence

How does concatenated SMS work?

  1. Decide the encoding

    If every character is in the GSM-7 alphabet (extension characters such as € [ ] { } ~ | count double), the text is GSM-7; otherwise the whole message becomes UCS-2.

  2. Check whether it fits one SMS

    Up to 160 GSM-7 septets or 70 UCS-2 code units travel as a single SMS with no header at all.

  3. Split into segments

    Longer text is cut into pieces of 153 septets or 67 code units. A careful encoder never splits an escape sequence or an emoji's surrogate pair across two parts.

  4. Add the header to each part

    Every segment carries the same reference, the total and its own sequence number. Over SMPP the UDH sits at the start of short_message with the UDHI bit set (esm_class 0x40); alternatively the sar_msg_ref_num, sar_total_segments and sar_segment_seqnum TLVs, or a single message_payload, let the SMSC do it.

  5. The phone reassembles

    The handset buffers parts by sender and reference until all have arrived, orders them by sequence number and shows one message. Each part produces its own delivery report.

Implementation

Concatenated SMS in code

import secrets
 
def concat_ucs2(text: str) -> list[bytes]:
    """Split text into UCS-2 SMS parts, each prefixed with a 6-octet concatenation UDH."""
    units = text.encode("utf-16-be")
    if len(units) <= 140:                    # up to 70 code units: one SMS, no header
        return [units]
    chunks, i = [], 0
    while i < len(units):
        end = min(i + 134, len(units))       # 67 code units per part
        # never cut between the two halves of a surrogate pair (an emoji)
        if end < len(units) and 0xD8 <= units[end - 2] <= 0xDB:
            end -= 2
        chunks.append(units[i:end])
        i = end
    ref, total = secrets.randbelow(256), len(chunks)   # same reference on every part
    return [bytes([0x05, 0x00, 0x03, ref, total, n + 1]) + c for n, c in enumerate(chunks)]
 
# Send each part as its own submit_sm with esm_class=0x40 (UDHI) and data_coding=0x08 (UCS-2).

Splitting on whole UTF-16 code units, and never inside a surrogate pair, keeps emoji and Indian scripts intact at part boundaries.

Failure modes

Common mistakes with concatenated SMS

  • Counting characters instead of segments

    A 161-character text is two SMS, not one and a bit. Compute segments with the encoding rules before every bulk send and show the count in the campaign tool.

  • Letting one character flip the encoding

    An emoji, a curly apostrophe pasted from a document or the rupee sign forces UCS-2 and can triple the segments. Normalise text, or choose UCS-2 deliberately for regional-language messages and budget 67 characters a part.

  • Splitting in the wrong place

    Cutting between an escape character and its partner, or between the halves of an emoji, corrupts the characters at the boundary. Split on whole characters, as the Python sample does.

  • Reusing reference numbers carelessly

    Two long messages to the same phone with the same reference can be merged into nonsense. Increment references per recipient and wrap at 255.

  • Reading one receipt as the whole message

    Each part has its own delivery report; the message is delivered only when every part is. Aggregate receipts under your own message ID before marking it delivered.

Example: A Surat wholesaler's weekly rate list

A Surat textile wholesaler sends a weekly rate update to 8,000 retailers. The English text is 290 GSM-7 characters: two segments, so 16,000 billed SMS. One week a staff member pastes the rates from a spreadsheet with curly quotes and a rupee sign, and the message silently becomes UCS-2: 290 characters now need five segments, or 40,000 billed SMS for the same news. The team adds a pre-send check that flags non-GSM characters, swaps curly quotes for straight ones and writes Rs. instead of the symbol, which brings the update back to two parts.

Recommended practice

Best practices for concatenated SMS

  • Keep OTPs and alerts to one segment: 160 GSM-7 or 70 UCS-2 characters.
  • Prefer message_payload or the SAR TLVs when your SMSC supports them rather than hand-building headers.
  • Test long messages on several handsets and networks before a campaign.
  • For long content, send a short SMS with a whitelisted link, or use WhatsApp, where a text message can run to 4,096 characters.

Design intent

Why does concatenated SMS matter?

Segment count is the hidden multiplier on every SMS bill: a campaign that spills one character past 160, or picks up a single emoji, can double or triple its cost overnight. It also affects reliability, because each extra part is another chance for something to go missing, and compliance in India, where the full text must still match a registered DLT content template however many segments it takes. Knowing the mechanics lets developers size messages before sending and explain to marketing why one more line has a price.

Concatenated SMS FAQs

Why does a long SMS hold 153 characters per part instead of 160?+
Because each part carries a 6-octet User Data Header that tells the phone how to rejoin the pieces. That leaves 134 octets for text: 153 GSM-7 characters or 67 UCS-2 characters.
Am I charged for each part of a concatenated SMS?+
Yes. Operators and providers bill every segment as a separate SMS, so a three-part message costs three SMS, and some providers bill on submission rather than on delivery.
What is the maximum length of a concatenated SMS?+
The header can number up to 255 parts, but providers, operators and handsets usually cap long messages far lower. Check your provider's limit and keep business texts to a few segments at most.
Why did my long SMS arrive in pieces or with strange characters?+
Usually because the header was lost or not recognised: a route that strips or mangles the UDH, or a phone that missed one part. The phone then shows the parts separately, or prints the header bytes as odd characters.

See InfiQ pricing

Put Concatenated SMS to work

InfiQ is an official Meta Business Partner. Start a 7-day free trial — no card needed — and see this running on your own number.

Start free trialBrowse all 550 terms