Arabic SMS Character Limit: 70, Then 67
Why an Arabic SMS holds 70 characters, why each part of a long one holds 67, how one Arabic letter flips a Latin text to UCS-2, and how to write short ones.
Risposta rapida
One Arabic SMS holds up to 70 characters. Arabic cannot be written in the GSM 7-bit alphabet, so it is sent as UCS-2, 16 bits per character, in a 140-octet payload (3GPP TS 23.038). Longer texts are split into parts, and each part holds 67 characters because 6 octets go on a concatenation header (TS 23.040). A single Arabic letter in an English text switches the whole message to UCS-2.
A reminder that looks short on screen can still be billed as two or three messages. If you send SMS in Arabic, or in a mix of Arabic and English, the cause is almost always the same: the character set. This guide explains where the numbers 70 and 67 come from, shows worked examples you can check yourself, and gives practical rules for writing Arabic texts that stay in one segment.
It is written for teams that send appointment reminders, payment links and one-time codes in the Gulf and the wider Arabic-speaking world, but the arithmetic is the same for any language written in a non-GSM script.
The short answer: 70 characters in one, 67 per part when split
| Message type | Single SMS | Each part of a long SMS |
|---|---|---|
| Latin text, GSM 7-bit default alphabet | 160 characters | 153 characters |
| Arabic (or any text with a non-GSM character), UCS-2 | 70 characters | 67 characters |
These figures come straight from the specifications: 160 and 70 from TS 23.038, and 153 and 67 from TS 23.040 (see Sources). So the rule of thumb for Arabic is: up to 70 characters is one SMS; from 71 characters it is two parts, from 135 three parts, and so on in steps of 67 (with the 8-bit reference header described below). The first extra part starts at character 71 because a message of 71 characters no longer fits in one 140-octet payload and must be split into parts that each lose room to the header.
Why: GSM-7 versus UCS-2 and the 140-octet payload
The text of an SMS travels in a field that holds 140 octets (8 bits each). What you can fit in those 140 octets depends on how each character is encoded.
- GSM 7-bit default alphabet. Each character takes 7 bits, so 140 octets hold (140 x 8) / 7 = 160 characters. TS 23.038 states this calculation directly.
- UCS-2. Each character takes 16 bits (2 octets), so 140 octets hold 70 characters. TS 23.038 says a UCS-2 message "can consist of up to 140 octets, i.e. up to 70 UCS2 characters".
The default alphabet is built around Latin letters, digits and common punctuation, plus a small extension table reached through an escape character. Arabic letters are not in it. In its language-indication table, TS 23.038 adds a note that message text in Hebrew, Arabic and Russian cannot be encoded in the GSM 7-bit default alphabet and that UCS-2 shall be used for these languages. So an Arabic SMS is a UCS-2 SMS.
The same specification also notes that when the extension table is used for a character, the number of displayable characters falls by one for each use. That is why a Latin message with several square brackets, curly braces or the euro sign can hold fewer than 160 characters. It is a smaller version of the same problem.
Where 67 comes from: the concatenation header
A message longer than one SMS is sent as several SMS parts (this is called concatenation). The phone needs to know that part 2 belongs with part 1, so each part begins with a small user data header. TS 23.040 describes the concatenated-message information element, with a reference number, the total number of parts and the sequence number of this part.
For the 8-bit reference number version, the header costs 6 octets of the 140. TS 23.040 then gives the three results in one place:
- 8-bit data: 134 (140 - 6) octets per part.
- GSM 7-bit default alphabet: 153 (160 - 7) characters per part. The specification states the figure; in short, the header takes seven 7-bit positions.
- UCS-2: 67 ((140 - 6) / 2) characters per part. A UCS-2 character must not be split in the middle.
TS 23.040 also defines a version of the header with a 16-bit reference number. It takes 7 octets, so each part holds 152 GSM-7 characters or 66 UCS-2 characters. Which version is used is up to the sending platform, so if your provider's counter shows 66 per part rather than 67, this is why.
The same clause sets the ceiling: at most 255 parts, which TS 23.040 gives as 17,085 UCS-2 characters for a fully concatenated message. In practice you should never come close, because providers usually bill each part as a separate message; check your own provider's pricing.
Esempio pratico (a titolo illustrativo)
Take an Arabic reminder of 107 characters, counting letters, spaces, digits and punctuation. It is over 70, so it is split. 107 / 67 = 1.6, so it takes 2 parts: the first holds 67 characters and the second holds the remaining 40. Two segments are billed. If the same reminder were trimmed to 70 characters or fewer, it would be one segment. That one edit can halve the cost of the send.
تذكير: موعدك في العيادة غدا الساعة 10:00 صباحا. للتأكيد ردوا بنعم، وللتغيير اتصلوا بنا. شكرا لكم على ثقتكم.
English note: "Reminder: your appointment at the clinic is tomorrow at 10:00 am. To confirm, reply yes; to change it, call us. Thank you for your trust." This sample has 107 characters, so it is sent in 2 parts (67 + 40).
تذكير: موعدك غدا الساعة 10 صباحا. للتأكيد ردوا بنعم.
English note: "Reminder: your appointment is tomorrow at 10 am. To confirm, reply yes." This version has 52 characters and fits in one SMS.
Mixed Arabic and English: one Arabic letter switches the whole text
An SMS carries one character set for the whole message. It does not switch between GSM-7 and UCS-2 inside the same text. So if even one character is outside the GSM 7-bit alphabet, the entire message is encoded as UCS-2 and every character, including all the English ones, now costs 16 bits.
Here is an illustrative example. This English reminder has 94 characters, all in the default alphabet, so it is one SMS:
Reminder: your appointment is tomorrow at 10:00. Reply YES to confirm or call us to change it. Add the Arabic word for "thank you" at the end, a nice local touch:
شكرا
The text is now 99 characters, and because it contains Arabic letters it is UCS-2. 99 is over 70, so it is split into two parts (67 + 32). One word turned a single-segment message into a two-segment one. The reverse also works: a short Arabic greeting followed by a long English body is a long UCS-2 message, so it is better to send one language per SMS, or keep the second language to a short, necessary phrase.
Be careful when you build SMS templates with variables. A customer's name is merged into the text at send time. If your template is pure English and the customer record holds a name written in Arabic script, that single name moves the whole message to UCS-2. Estimate cost with the longest realistic name, not with a Latin placeholder.
Hidden characters that cost you
These do not look like extra letters, but each one is a character in the count, and some push a Latin message out of GSM-7.
- Diacritics (tashkeel). The Unicode Arabic block includes marks such as fathatan (U+064B) and shadda (U+0651) as separate characters. Each mark adds to the count, so a fully vowelled word is much longer than the same word written plain. For SMS, leave them out unless they change the meaning.
- Tatweel. The elongation character (U+0640) is a character like any other. Do not use it to stretch words.
- Smart quotes and typographic apostrophes. A curly apostrophe copied from a word processor is not in the default alphabet, so it can flip an otherwise plain English message to UCS-2. Type a straight apostrophe in templates.
- Emoji. They are not in the default alphabet, so they force UCS-2. Your platform's counter will show the true size.
- Invisible direction marks. Text pasted from a document can carry invisible right-to-left or left-to-right control characters. The Unicode Arabic chart lists an Arabic Letter Mark (U+061C), for example. They are invisible on screen but still characters in the count and still outside GSM-7.
- Arabic-Indic digits. Digits in the Arabic-Indic block (U+0660 to U+0669) are different characters from 0 to 9. In a mostly Arabic message it makes no difference to the encoding, but in an English message they would force UCS-2. For codes, use the plain 0 to 9 so that every system reads the same value.
Writing short Arabic reminders, payment links and OTP texts
The goal is under 70 characters wherever possible. The following rules usually get you there without making the message curt.
- Lead with the one fact the reader needs. Date and time first, then the action.
- Cut greetings that repeat what the sender ID already says. If the sender ID shows your brand name, you do not need to start with it.
- Use a short link. A 30-character URL takes over 40 percent of a single UCS-2 segment. Use your own short domain and put the link last.
- Write one action per message. "Reply yes" or "tap the link", not both.
- Check the count with a real tool before a campaign, using a sample with the longest name and the longest date format you will merge in.
Illustrative examples, with character counts (spaces and punctuation included):
رمز التحقق: 482913. لا تشاركه مع أحد.
English note: "Verification code: 482913. Do not share it with anyone." This one-time code text has 37 characters and fits in one SMS with room to spare.
فاتورتك جاهزة. ادفع من هنا:
https://example.com/p/8F3K
English note: "Your invoice is ready. Pay here: link." This payment text has 54 characters, including a 26-character example link, and stays in one segment. Notice that the link characters are Latin, but because the message contains Arabic, they still cost 16 bits each.
تذكير: موعدك غدا الساعة 10 صباحا. شكرا لك.
English note: "Reminder: your appointment is tomorrow at 10 am. Thank you." This reminder has 42 characters when you count the spaces and punctuation, which leaves space for a short link or a reply keyword. For appointment flows, see our guide to promemoria degli appuntamenti.
A one-time code text should contain the code and the warning, nothing else. Do not add marketing lines to a code message.
SMS or WhatsApp for long Arabic messages
SMS is a good fit for something short and urgent: a code, a time, a payment link. When the content is long, such as a full set of instructions or a list of options, an SMS split into four or five parts is expensive. A common pattern is to send a short SMS with a link, or to use WhatsApp where the customer has opted in and your message fits the channel's template rules. WhatsApp is a separate channel with its own pricing, which we cover in WhatsApp Business API pricing, and it does not count in SMS segments.
The right mix depends on your consent records, your provider and your audience. Our guide to WhatsApp and SMS follow-up for outbound teams covers how to sequence the two after a call, and missed call recovery for Gulf businesses shows where a short Arabic text fits in a follow-up routine.
Come si inserisce DialerBee
Sul WhatsApp e SMS page, DialerBee states that its SMS segmentation is character-set aware: GSM-7 and UCS-2 are counted differently, and an Arabic message is counted in UCS-2, so the segment count shown is the one used for billing. The same page lists SMS sender IDs, templates held in a shared library, delivery reports per message and a balance visible in the console. Appointment reminders and published surveys can go out over SMS and WhatsApp, and in the casella di posta omnicanale voice, WhatsApp, SMS and email threads sit in one conversation list.
Whichever platform you use, keep the check simple: write the template, merge a long realistic name, read the segment count, and shorten until you are under 70 characters.
Domande frequenti
How many characters fit in one Arabic SMS?
Up to 70. Arabic letters cannot be stored in the GSM 7-bit default alphabet, so the message is sent as UCS-2, which uses 16 bits per character. TS 23.038 says a UCS-2 message can use up to 140 octets, i.e. up to 70 characters. A Latin-only message in the default alphabet fits up to 160 characters in one SMS.
Why does a long Arabic SMS hold only 67 characters per part?
Once a message is split, each part has to carry a concatenation header so the phone can put the parts back in order. For the version with an 8-bit reference number, TS 23.040 subtracts 6 octets for this header, which leaves 134 of the 140 octets for text. At 2 octets per UCS-2 character, that is 67 characters per part. If the sender uses the 16-bit reference version instead, the header takes 7 octets and each part holds 66 characters.
Why was my Arabic SMS split into several messages?
Because it was longer than 70 characters. As soon as the text is over 70 UCS-2 characters, it is sent as a multi-part message of up to 67 characters per part. Count the characters, not the words, and remember that spaces, digits, punctuation and links all count.
Does one Arabic word in an English SMS change how it is sent?
Yes. An SMS has one character set for the whole message. If any character falls outside the GSM 7-bit alphabet, the whole text is sent as UCS-2, and the limit drops from 160 to 70 characters (67 per part when split). In the worked example in this guide, a 94-character English reminder fits in one SMS, but the same text plus one Arabic word is 99 UCS-2 characters and takes two.
Do Arabic diacritics and emoji count as characters?
Diacritics such as the fatha or shadda are separate Unicode characters, so each one is counted on its own. Emoji and curly quotation marks are not in the default alphabet, so they also force UCS-2. Your SMS platform's own segment counter is the safest check, because it counts exactly what will be sent.
Is WhatsApp a way around the SMS limit for long Arabic messages?
WhatsApp is a different channel with its own rules, and it is not counted in SMS segments. One option is to use SMS for a short alert with a link and WhatsApp for longer content. Which channel is allowed for a given message depends on your consent records and your provider's template rules, so check those before you switch.
Fonti
- 3GPP TS 23.038, Alphabets and language-specific information (version 19.0.0, Release 19, ETSI TS 123 038 V19.0.0, October 2025): GSM 7-bit default alphabet up to 160 characters, UCS-2 up to 140 octets or 70 characters, and the Arabic UCS-2 note. etsi.org/deliver/etsi_ts/123000_123099/123038/ E 3gpp.org/DynaReport/23038.htm (checked November 7, 2026)
- 3GPP TS 23.040, Technical realization of the Short Message Service (SMS) (version 19.0.0, Release 19, ETSI TS 123 040 V19.0.0, October 2025): user data of 140 octets, concatenated short messages with 6-octet header, 153 and 67 characters per part (clause 9.2.3.24.1), and 152 and 66 with the 16-bit reference number (clause 9.2.3.24.8). etsi.org/deliver/etsi_ts/123000_123099/123040/ E 3gpp.org/DynaReport/23040.htm (checked November 7, 2026)
- The Unicode Standard, Arabic block chart (U+0600 to U+06FF): tatweel, fathatan, shadda, Arabic Letter Mark and Arabic-Indic digits. unicode.org/charts/PDF/U0600.pdf (checked November 7, 2026)
Character counts in this guide are illustrative and were counted by hand in code points. Always confirm segment counts and billing with your own SMS provider, because carriers and platforms can apply their own handling.
Letture correlate
Articoli correlati
Arabic Speech Recognition Across Dialects in Call Centres
10 minuti di lettura
TecnologiaArabic TTS for Numbers, Dates and Prices in Voice Agents
10 minuti di lettura
TecnologiaWriting Prompts for an Arabic Voice Agent
11 minuti di lettura
GuideArabic Voice Agent Testing Checklist Before You Go Live
Tempo di lettura: 12 minuti
Pronti a vedere DialerBee in azione?
Richiedi una demo dal vivo di 15 minuti oppure avvia una prova gratuita di 14 giorni: niente slide, nessun impegno. Le chiamate partono quando il tuo trunk SIP o PBX è collegato e approvato.
Prova gratuita di 14 giorni · credito IA gratuito · senza carta · 11 lingue · BYOC · controlli a supporto della conformità