Reading PDF417 Barcodes on North American Driver Licenses

If you have ever verified a US or Canadian driver license programmatically, you have run into PDF417. It is the dense, rectangular barcode printed on the back of nearly every North American license and state ID. Understanding what it contains — and why it matters — makes your verification flow faster and more reliable.

What is PDF417?

PDF417 is a two-dimensional stacked barcode format. Unlike a QR code, it is wider than it is tall and encodes data in rows of stacked linear patterns. The name comes from its structure: each pattern is made of 4 bars and spaces spanning 17 units.

The key advantage for identity documents is capacity. A single PDF417 symbol can hold hundreds of bytes of text, which is more than enough for a full set of licence holder fields. That is why the standard-setting body for North American licenses adopted it as the machine-readable format on the back of the card.

Why it exists on driver licenses

The barcode duplicates the human-readable data printed on the front of the license. A police officer, bartender, or verification system can scan the back and instantly read structured fields instead of retyping them. For automated KYC, this matters because barcode data is machine-formatted and does not suffer from the OCR ambiguities you sometimes see reading printed text (think 0 vs O, or worn fonts).

What data lives inside the barcode

The AAMVA (American Association of Motor Vehicle Administrators) standard defines the field layout. Common elements include:

• First, middle, and last name
• Date of birth
• Full address
• License or ID number
• Issue and expiry dates
• Document discriminator (a unique per-card value)
• Sex, height, eye color, and hair color where issued
• Issuing jurisdiction and AAMVA version

Fields are identified by short element IDs. For example, DAC is the first name, DCS is the family name, DBB is the date of birth, and DAQ is the license number. Each jurisdiction and AAMVA version can vary slightly in which optional fields are present and how they are formatted, so a robust parser has to account for multiple standard revisions.

Why barcode reading strengthens verification

Reading the barcode is not just a convenience. It gives you a second, independent source of the same identity data — and that opens up meaningful fraud checks.

Cross-validation against the printed front

A genuine license should have matching data on the printed front and the encoded back. If someone alters the printed name or date of birth but leaves the barcode untouched (a common shortcut in casual fraud), a front-versus-barcode comparison exposes the mismatch immediately. This is one of the most reliable low-cost signals in North American document checks.

Cleaner structured data

Barcode fields are already delimited and typed. You do not have to guess where a name ends or normalize a date format printed in an unusual style. For downstream systems — onboarding forms, age gates, AML screening — clean structured input reduces errors and manual review.

Faster, more consistent extraction

Because the barcode is machine-formatted, extraction is deterministic. There is less variance across lighting conditions and image quality than with pure OCR of the printed front, so your automated pass rate tends to be higher when both sides are captured.

Practical capture tips

PDF417 is dense, so image quality drives success. A few things that consistently help:

• Capture the back of the card. The barcode is not on the front. Prompt users clearly to flip the document.
• Ask for adequate resolution. The fine bar-and-space pattern needs enough pixels to resolve. Very small or blurry images fail to decode.
• Avoid glare and folds. Reflections across the barcode break the symbol. Matte, evenly lit captures decode best.
• Handle both sides in your flow. For a full check, capture the front for OCR, portrait, and authentication, and the back for the barcode.

What to do with a decode failure

Not every image decodes on the first try — worn cards, damage, or poor capture happen. Build a retake prompt into your flow rather than hard-failing the user. When the barcode still will not read, you can fall back to front-side OCR, but you lose the cross-validation signal, so treat those cases with slightly more scrutiny.

Fitting barcode reading into a full check

Barcode parsing is one layer. A complete North American identity check usually combines several capabilities: OCR of the front, MRZ or barcode reading, document authentication to catch forgeries and tampering, and — for higher assurance — a biometric face match against a selfie with liveness detection.

ID Analyzer supports document OCR and MRZ/barcode reading across a wide range of North American licenses and IDs, alongside document authentication and biometric verification, so you can assemble the exact combination your risk model needs.

Takeaways

PDF417 is the workhorse of North American license data. It carries a structured copy of the license holder details, decodes deterministically, and — most importantly — gives you an independent source to cross-check against the printed front. If you verify US or Canadian licenses and you are only reading the front, you are leaving a valuable fraud signal on the table. Capture both sides, parse the barcode, and compare.