Sophie
6 November 2026 · 12 min read
You're writing a short message to a loved one in Morocco, Spain or Quebec. Two sentences, a heart at the end, that's it. And your bill shows three international text messages instead of one. A standard SMS holds 160 characters only if it uses nothing but the 7-bit GSM alphabet. The moment a single character falls outside that table — an emoji, an "œ", a curly quotation mark pasted in by a copy-paste, an Arabic, Cyrillic or Chinese letter — the entire message switches to Unicode (UCS-2) and the limit drops to 70 characters. Beyond that, it is split into segments, each billed at the price of a full SMS. Sent abroad, where a segment often costs between €0.10 and €0.50, a 150-character message with one emoji costs three times the advertised price.
Updated 6 November 2026.
Why does an SMS go from 160 to 70 characters?
Because an SMS carries 140 bytes, not 160 characters. With the 7-bit GSM alphabet, each character weighs 7 bits: 140 bytes = 1,120 bits = 160 characters. In Unicode, each character weighs 16 bits: 140 bytes = 70 characters. The container never changes; what changes is the encoding.
This mechanism comes from the GSM 03.38 standard, now maintained by 3GPP under the reference TS 23.038. It defines a table of 128 so-called "basic" characters — unaccented letters, digits, common punctuation — plus an extension table of a few symbols ({, }, [, ], \, ~, ^, |, €) that count double. We covered the origin of this constraint in our article on the 160-character limit of an SMS.
The point almost nobody knows: the 7-bit GSM table accepts only a handful of accents. It includes à, è, é, ù, ì, ò, ç, ä, ö, ü, ñ, å, æ, ø. It does not include, among others:
| Character typed | In the GSM table? | Consequence |
|---|---|---|
é, è, à, ç, ü, ñ | Yes | 160 characters retained |
ê, î, û, â, ô, ï | No | Switches to Unicode → 70 |
œ, Œ, uppercase æ | No (except lowercase æ) | Switches to Unicode → 70 |
’ (curly apostrophe) | No | Switches to Unicode → 70 |
« », —, … | No | Switches to Unicode → 70 |
| Emoji (😀, ❤️, 👍) | No | Switches to Unicode → 70 |
| Arabic, Cyrillic, Greek, Chinese | No | Switches to Unicode → 70 |
In other words: writing à bientôt ("see you soon") costs you 90 characters because of the single circumflex on the "ô". Writing à bientot ("see you soon", without the circumflex) keeps them. It's absurd, but it has been the standard since 1987.
What is a "concatenated" SMS and why does it cost more?
A message that is too long is not truncated: it is cut into several text messages sent separately, then reassembled on arrival. Your recipient sees a single message; the network, for its part, carried two, three or ten — and bills for all of them.
This splitting is called concatenation ("multipart" or "long" SMS). Each segment has to carry a technical header (UDH) indicating the segment number and the total, which eats into the usable space. The actual capacity per segment therefore drops:
| Message type | 7-bit GSM encoding | Unicode encoding |
|---|---|---|
| Single SMS | 160 characters | 70 characters |
| 2 segments | 306 (2 × 153) | 134 (2 × 67) |
| 3 segments | 459 (3 × 153) | 201 (3 × 67) |
| 5 segments | 765 | 335 |
The shift from 160 to 153 (and from 70 to 67) is not a mistake: those are the characters lost to the reassembly header as soon as the message runs to more than one segment.
The impact on price is direct. A 150-character message with no emoji = 1 SMS. The same message with an emoji = 3 segments (150 > 134). To a destination billed outside your plan, that means three times the unit price. Within the European Union, where roaming regulated by the European Commission applies your included texts, it uses three units of your allowance instead of one.
Does this affect the texts included in my plan?
Yes, but invisibly. Almost all consumer plans in France include "unlimited texts" within mainland France: in that case, splitting changes nothing on your bill. It's internationally and when roaming outside the EU that the mechanism becomes expensive.
Three situations to distinguish:
- A text to a French number, from France: generally unlimited with Orange, SFR, Bouygues Telecom and Free Mobile. The number of segments makes no difference to you.
- A text sent from France to a foreign number: this is the trickiest case. With most operators, international texts are not included in the unlimited allowance, or are included only for a restricted list of destinations. Every segment is billed.
- A text sent from abroad (roaming): within the EU and the EEA, your texts are counted as if you were in France; outside the EU, they are billed per unit — and therefore per segment. We covered this in our guide to texts and data outside the European Union.
A useful habit before you leave: check your operator's official price sheet (the mandatory tariff information document) for the price of a text to the zone concerned, then mentally multiply it by three if you write with emojis. Arcep's tables on international roaming are a reminder that pricing outside the EU remains freely set by operators.
How can I tell how many characters I'm actually using?
No mainstream messaging app shows it clearly by default. You have to switch on a counter, or work it out by deduction.
On Android with Google Messages (2026 version), the path is: Google Messages → profile picture → Message settings → Advanced → Split long SMS. Turn the option on and the character counter appears at the end of the input field as soon as the message approaches the limit. It shows a format such as 12/2: 12 characters left in the 2nd segment.
On iPhone (iOS 18 and later), Apple removed the character counter from Settings → Messages several iOS generations ago. There is no longer a reliable native display. Two workarounds:
- Type the message in the Notes app, which imposes no limit, and check its length before pasting it.
- Use the Shortcuts app to create a "Count Characters in Text" action and run it from the share sheet.
The simplest test remains empirical: write your message, delete all emojis and replace curly apostrophes with straight ones, then see whether the counter jumps. If it goes from 70 to 160, you've just saved two-thirds of the cost.
Which characters should I replace to stay at 160?
The culprits are almost always the same, and they often turn up without you typing them: pasted in from a website, an email or a word processor.
The hall of fame of invisible characters that flip a message:
- The curly apostrophe
’instead of the straight apostrophe'. It's the number one trap in French: the autocorrect on many keyboards, and copying from a website, insert it without warning. A single’in l'adresse ("the address") and your 140-character message becomes two text messages. - French guillemets
«»and curly quotation marks"", often pasted in from an article. - The single-character ellipsis
…instead of three full stops. - The em dash
—inserted automatically in place of a hyphen. - The non-breaking space, invisible to the eye, generated by some word processors before a semicolon or an exclamation mark.
- Circumflexes and diaereses other than those listed above:
ê,î,û,â,ô,ï,ë.
One effective habit: never draft an international text message in a word processor. If you prepare long or recurring messages, a plain paper notebook or a plain-text editor avoids automatic typographic substitutions. And if you type a lot on a phone, a compact Bluetooth keyboard makes it easier to proofread before sending — it's a trivial point, but that's often where the stray character slips in.
Rule of thumb: if your message absolutely has to fit into a single SMS sent abroad, aim for 60 characters when it contains an emoji, and 150 when it contains none. The margin absorbs surprises.
Non-Latin alphabets: why it's even worse
A message in Arabic, Cyrillic, Greek, Hebrew, Thai or Chinese is always in Unicode. There is no "7-bit" version of these alphabets in the base standard.
The consequence: a text written in Arabic to the Maghreb or in Cyrillic to Eastern Europe is capped at 70 characters from the very first word. This matters particularly for families who write in their original language: three sentences in Arabic easily run to four or five segments. Our article on texts to Africa and the Maghreb sets out typical prices by destination.
Two technical nuances exist but don't concern you directly:
- Some networks support national substitution tables (National Language Shift Tables, 3GPP TS 23.038) for Turkish, Spanish, Portuguese and a few others. These allow roughly 155 characters to be kept instead of 70. Whether they work depends on the handset and on the receiving network: you can't rely on them internationally.
- Professional sending platforms (A2P SMS) bill explicitly per segment and display the encoding they detect. That's why business text messages systematically avoid emojis.
What about RCS or a messaging app?
The character limit disappears: RCS and messaging apps run over data and have no notion of a 140-byte segment.
It's the simplest solution when the message is long or full of emojis. But it comes with a strict condition: both correspondents need a data connection and a compatible app. RCS, described in our comparison RCS versus SMS, automatically falls back to SMS when data is unavailable — and the long message is then split and billed again.
For a trip, the choice is made in advance:
| Need | Best option | Limitation |
|---|---|---|
| Short, urgent message, no data | SMS in strict ASCII, 150 characters | No emoji, no exotic accents |
| Long message, data available | RCS or a messaging app | Requires data on both sides |
| Recipient without a smartphone | SMS is the only option | Count the segments |
| Area with no network | Satellite messaging | Very short messages, check operator offers |
In a country where data is expensive or patchy, a 10,000 mAh power bank is more useful than any attempt to optimise encoding: what usually stops a message getting through is a flat battery. And for anyone leaving for a long stretch, a multi-port USB-C charger saves you having to choose between the phone and everything else.
Three checks before sending an international text
- Reread the punctuation. Replace
’with',…with...,«»with". Delete the emojis if the message exceeds 70 characters. - Check the number format. A perfectly encoded message to a badly written number won't go anywhere:
+followed by the country code, without the national0. We go through the various cases in our guide to the international format of a phone number. - Check the price per segment. On your operator's price sheet, for the zone concerned. Then multiply by the estimated number of segments.
If you regularly send messages abroad and the segments add up, our SMS sending service displays the split directly, and the pricing page breaks down the cost by destination. For recurring questions about encoding, delivery receipts and undelivered messages, the FAQ gathers the technical answers.
Frequently asked questions
How many characters does an emoji count for in a text message? Technically, a simple emoji takes up 2 Unicode characters (4 for composed emojis such as flags or families). But its size isn't the main effect: it's that it switches the whole message to Unicode, bringing the limit down from 160 to 70 characters.
Why does my text arrive in several pieces on my recipient's phone? Because reassembly failed. Segments are put back together by the receiving handset using the UDH header; if they arrive out of order, too far apart in time, or if the phone is very old, they show up as separate messages.
Do French accents really switch a text message to Unicode?
Only some of them. é, è, à, ç, ù are in the GSM table and cost nothing. ê, î, û, â, ô, ï, ë and œ are not, and they drop the limit to 70 characters.
Do all operators bill for the split? Yes, the principle is universal: billing is per text message carried, and therefore per segment. What varies is the unit price depending on the destination and your plan — and the fact that texts to mainland France are generally unlimited, which hides the phenomenon entirely.
How can I write in Arabic or Cyrillic without blowing up my text message count? There is no fix on the SMS side: these alphabets are always Unicode, so 70 characters per segment. For long messages, RCS or a messaging app over data is the only economical option.
