Code 128 Explained: Subsets, FNC1 and Real-World Use

The short answer

Code 128 is a 1D barcode that encodes all 128 ASCII characters, used almost exclusively in logistics and asset tracking rather than retail POS. It picks between three internal subsets (A, B and C) to keep the symbol as narrow as possible. When the first character is an FNC1, the same barcode is called GS1-128 and the payload is read as GS1 Application Identifiers — GTIN, batch, expiry, serial, pallet number.

Code 128 at a glance

Where Code 128 fits depends on what else you considered. Most teams pick between Code 128, Code 39 and GS1-128 — they look similar at a glance and the wrong choice surfaces months later when the data won't parse cleanly.

The three subsets: A, B and C

The clever part of Code 128 is that it doesn't encode characters one at a time. Instead it carries an internal state that selects one of three subsets and switches between them mid-symbol with a single shift character. The encoder picks whichever subset makes the next few characters cheapest, and the decoded text comes out as a clean ASCII string with no trace of the underlying gymnastics.

• Subset A — ASCII 0 through 95: uppercase letters, digits and control codes (NUL, GS, RS, FS and the rest). Used mainly in legacy systems that need to embed control characters in the payload.
• Subset B — ASCII 32 through 127: uppercase, lowercase, digits and punctuation. The default for general-purpose alphanumeric text — "Code 128 supports lowercase" means Subset B.
• Subset C — pairs of digits from 00 to 99 encoded as one symbol character. The killer feature for pure numeric payloads: a 12-digit value takes six Subset C characters instead of twelve Subset B characters, halving the printed symbol width.

You never have to think about subsets when decoding — the scanner returns plain ASCII either way. The reason to know about them is when you're building a Code 128 from scratch and the symbol comes out wider than expected: usually a numeric run that the encoder failed to fold into Subset C.

FNC1 and GS1-128

FNC1 is the function character that turns a plain Code 128 into a GS1-128. It plays two roles:

1. Mode flag. An FNC1 immediately after the Code 128 start character marks the entire symbol as a GS1 carrier. Without that flag the payload is read as free-form ASCII; with it, parsers expect GS1 Application Identifiers.
2. Variable-length separator. Some AIs are fixed length (GTIN is always 14 digits inside AI 01; the expiry date inside AI 17 is always 6 digits). Others — batch (10), serial (21), customer reference (240) — are variable up to their maximum. The encoder appends an FNC1 immediately after a variable-length AI so the parser knows the field has ended. Fixed-length AIs don't need a trailing FNC1 because their length is known from the AI itself.

A worked GS1-128 example

Take the worked example used throughout the GS1 logistics guides: (01)07640156790009(10)ABC123(17)261231. It carries a GTIN, a batch number and an expiry date in a single GS1-128 symbol.

Reading the symbol left-to-right, the parser walks through it like this:

1. Start C + FNC1. The encoder chose Subset C because the payload opens with a long numeric run. The FNC1 immediately after the start character marks the symbol as GS1-128.
2. AI (01) — GTIN-14. Fixed length: 14 digits. 07640156790009 is the product identifier. No trailing FNC1 needed because GS1 knows AI 01 is exactly 14 digits.
3. AI (10) — Batch. Variable length, up to 20 alphanumeric characters. The encoder switched out of Subset C into Subset B for ABC123, then appended an FNC1 to mark the end of the variable field.
4. AI (17) — Expiry. Fixed length: 6 digits (YYMMDD). 261231 means 31 December 2026.
5. Check digit + stop. Code 128 finishes with a modulo-103 checksum and a stop pattern, both invisible in the decoded ASCII.

Paste the same value into the decoder and you can see each AI laid out with its spec name and validation status:

Where you actually see it

Code 128 is the dominant 1D symbology of the back-of-house. You meet it everywhere a workflow needs to encode more than a GTIN but doesn't justify a 2D code:

• Shipping labels. Tracking numbers from FedEx, UPS, DHL and Royal Mail are almost always Code 128 — alphanumeric, predictable length, easy to print on a thermal label.
• Pallet and case tags. The GS1 logistics label uses GS1-128 to carry the SSCC (AI 00) and, where required, the GTIN of the case contents (AI 02) plus batch and expiry.
• Asset and inventory stickers. Internal labels for equipment, shelves and reusable containers — "ASSET-2026-00042" style — sit on plain Code 128 with no GS1 layer.
• Healthcare wristbands. Patient ID bands often use Code 128 for its lowercase support and its ability to fit twenty-plus characters on a 10 mm-tall strip.
• Library and archive catalogues. Some legacy systems still emit Code 39 here; modern ones moved to Code 128 for the density.

To roll out new labels yourself, the Barcode Generator covers plain Code 128 with the encoding controls exposed; the ITF-14 Carton Label Generator is the right tool when the carton-level GTIN is the main subject.

Decoding Code 128 in software

The most common Code 128 bug isn't in the barcode — it's in the configuration between the scanner and the application that receives the data. Three settings cover most of the failures:

1. FNC1 transmission. Configure the scanner to emit FNC1 as ASCII 29 (GS) — the GS1 default. Some scanners are set to emit FNC1 as a blank character, which is what destroys the example at the top of this guide.
2. Prefix / suffix. Many scanners append a carriage return or tab on every read. That's fine for keyboard-wedge workflows; less fine when you're piping the raw payload into a regex.
3. Code-set hint. A handful of scanners can be asked to prefix every Code 128 read with ]C1 to signal a GS1-128 even when the payload doesn't open with FNC1. Useful for catching mis-formatted symbols upstream.

If you have a payload but aren't sure whether it's Code 128, Code 39 or something else entirely, the Barcode Type Detector will tell you in one step:

For round-trip QA — scanning your own printed Code 128 to verify it decodes cleanly — use the Online Barcode Reader from a phone or webcam. For bulk validation across hundreds of historical payloads, paste the list into the Batch Barcode Tools.

Common mistakes

The repeat offenders that turn a perfectly good Code 128 into a downstream parsing headache:

• Splitting GS1-128 AIs on a hard-coded character. Variable-length fields end up concatenated with the next AI when FNC1 is stripped. Always split on the configured FNC1 replacement (typically ASCII 29).
• Sending mixed-case text through a Subset A encoder. Lowercase letters silently disappear because Subset A doesn't cover them. Use Subset B (or let the encoder pick).
• Printing too narrow. Code 128 needs a minimum X-dimension (single-module width) to print and scan reliably. Shrink below it for label real-estate reasons and the symbol grade collapses on the first scratch.
• Confusing Code 128 with Code 39. Code 39 can't encode lowercase and is roughly 2× wider for the same payload. If your "Code 128" looks twice as long as the same data printed elsewhere, it's probably Code 39.
• Assuming GS1-128 implies a specific AI order. AIs can appear in any order — what matters is the FNC1 after every variable-length one. Reorder them all you like during artwork QA; the parser doesn't care.

Quick decision guide

When you only need the practical answer for a new label, the choices collapse to four short ones:

• Choose Code 128 when the payload is alphanumeric, you don't need GS1 structured data, and the label has room for a 1D symbol — internal asset tags, shipping labels, asset stickers.
• Choose GS1-128 when the payload carries GS1 Application Identifiers — SSCC pallet IDs, GTIN+batch+expiry on cartons, traceability data inside the logistics label.
• Avoid Code 39 unless you're forced to by a legacy reader. It's wider than Code 128 for the same data and can't carry lowercase.
• Move to 2D (Data Matrix or QR) when the payload pushes past ~40 characters, when the label is square rather than wide, or when you want the Sunrise-2027-ready GS1 Digital Link form on the consumer pack.

References

• ISO/IEC 15417:2007 — the international standard that defines the Code 128 symbology (subsets, FNC1, checksum algorithm). Available from the ISO catalogue.
• GS1 General Specifications — defines GS1-128, the Application Identifier table and the parsing rules. Published by GS1 and updated annually.
• GS1 Logistics Label guide — the practical reference for SSCC, AI ordering and label layout on cases and pallets: gs1.org/standards/barcodes/gs1-128.