Free · Intermediate · ~9 min read

Scheme Differences: Why "the Same" ISO 8583 Parses Differently

ISO 8583 is a standard, yet each scheme (Visa / Mastercard / UnionPay …) has its own conventions for encoding, bitmaps, headers, and private data elements. Parsing every scheme with one recipe usually misaligns. Here's what differs and how to map it.

1. Field encoding: BCD vs EBCDIC vs ASCII

The same numeric field or length prefix may use different encodings across schemes: packed BCD (two digits per byte), EBCDIC (IBM mainframe charset), or ASCII. A single message often mixes them — numeric in BCD, text in EBCDIC, chip data in binary. This is the classic beginner trap: decoding the whole message with one encoding inevitably misaligns.

2. Message headers

Some schemes prepend a self-describing header before the MTI (first byte = header length) carrying total length, format flags, etc.; others have no header and start straight at the MTI. Decide "strip a header or not, and how many bytes" first, or the MTI and bitmap shift as a block.

3. Bitmaps: two or three

The standard is a primary plus optional secondary bitmap (128 DEs). Some schemes support a third bitmap (DEs 129–192) for extra private fields. Check bit 1 and bit 65 to know how many bitmaps follow.

4. Private data elements: DE48 / 62 / 63 / 55

DE	Typical use	Format difference
DE48	Additional private data	Some use PDS (tag+len+value subfields)
DE55	EMV chip data	All BER-TLV, but length-prefix encoding/size varies
DE62 / 63	Switching / network private data	Often carry scheme-specific sub-bitmaps; fully private layouts

The internal subfield layouts of DE48/62/63 are closed, scheme-specific specs; without the relevant docs they're hard to decode exactly. Generic parsers usually surface them as raw data.

5. Practical advice

Before parsing, pin down scheme + header-or-not + overall encoding convention, then read each DE by its format.
Numeric → BCD, text → EBCDIC, DE55 → binary TLV — the most common mixed convention.
When unsure, let a tool auto-detect the encoding, then sanity-check.

👉 The parser tool here supports per-scheme parsing + auto encoding detection. Related: ISO 8583 basics, EMV glossary.