Technical guide

X9.37, ACH, and Cheque Processing: What Each Does

X9.37 and ACH are often mentioned together, but they solve different jobs. One moves cheque images and MICR data between institutions; the other moves batch electronic debit and credit entries. This guide explains the difference, the data each carries, and the checklist for connecting a cheque workflow to either rail.

Last reviewed 2026-10-20.

The short version

QuestionX9.37 / image cash letterNACHA ACH
What movesCheque images plus MICR and cash-letter dataBatch debit/credit entries and payment data
Carries an image?Yes — front and usually backNo
Typical useCheque image exchange and clearingElectronic payments, conversions, credits/debits
Key identifiersBundle, sequence, routing, account, cheque numberCompany ID, SEC code, trace number, routing/account
Return pathImage return / return cash letterACH return codes and notifications of change
Operational ownerClearing/image exchange teamACH/payments operations

What X9.37 is

X9.37 is an image exchange format used in cheque clearing. A file contains a cash letter with one or more bundles and items. Each item carries the cheque image, MICR line, amount, and sequence information so the receiving institution can process it as a cheque. The format is associated with Check 21 image exchange in the United States, but the underlying idea — exchange the image and data instead of the paper — is used in other image-clearing environments too.

The important implementation detail is not the file extension. It is the control total: the item count and amount totals in the file must reconcile to the bundle and cash letter. A file that parses but does not balance is not a successful clearing file.

What ACH is

ACH is a batch electronic funds transfer network. An entry identifies the receiver, the account, the amount, the effective date, and a SEC code that describes the payment context. Cheque-related ACH entries commonly appear when a cheque is converted at the point of purchase or via accounts receivable conversion, but the ACH entry is not a cheque image and does not carry the cheque’s front and back.

ACH has its own return process, notification-of-change rules, and settlement windows. Treat ACH origination as a separate integration with its own compliance, authorization, and reconciliation requirements — not as a file export from an OCR workflow.

Where they meet in a cheque workflow

  1. Capture the cheque image and MICR data.
  2. Extract and validate fields, including CAR/LAR agreement and date rules.
  3. Route exceptions to review; record the decision and evidence.
  4. Create the clearing artifact (X9.37, ACH, or another rail) through the bank’s system.
  5. Reconcile the item to the deposit, ledger, return, and settlement records.

The OCR/extraction layer is step 1–3. The clearing artifact is step 4 and belongs to a different system or a separately scoped integration. Do not collapse those boundaries in the architecture or the audit trail.

Integration checklist

  • Field mapping: routing, account, cheque number, amount, date, payee, item reference, and the bank’s own IDs.
  • Control totals: item count and amount totals reconcile at item, bundle, and file level.
  • Idempotency: the same cheque image cannot create two clearing entries.
  • Duplicate detection: check number, account, amount, date, and image similarity — not just a single field.
  • Returns: map return codes back to the original item and reviewer decision.
  • Audit trail: image, extracted fields, corrections, approvals, file ID, transmission response, and settlement reference.
  • Retention: align image, file, and audit retention with bank policy and legal hold.

How Chequedb fits

Chequedb focuses on capture-adjacent extraction, field validation, review, approvals, and audit evidence. It can produce structured fields and reviewable records that feed a bank’s existing X9.37 or ACH pipeline. X9.37 generation, ACH origination, transmission, and clearing are deployment- and integration-specific; confirm the current scope before assuming the OCR response is a cash letter or an ACH file.

The API contract is documented in the bank check OCR API docs. The security and data-flow boundary is covered in the security and compliance hub.

Common mistakes

  • Treating OCR success as deposit or settlement.
  • Letting the CAR/LAR mismatch pass because both fields were “read.”
  • Creating a file without reconciling control totals.
  • Using one field for duplicate detection and missing the same cheque in a different channel.
  • Losing the link between the return, the original image, and the reviewer decision.
  • Assuming an ACH entry and an X9.37 item are interchangeable because both came from a cheque.

Evaluate this on your own cheques

Bring representative de-identified images, target fields, and the downstream workflow. The evaluation should include failures, abstentions, review effort, and the integration boundary — not only a clean demo.

Related pages

Common questions

What is an X9.37 file?

X9.37 is an image exchange format used for cheque clearing. It carries a cash letter structure with cheque images, MICR data, bundle and sequence information, and forward or return records. Banks and image exchange networks use it to move cheque images and associated data between institutions.

What is the difference between X9.37 and ACH?

X9.37 is an image-based cheque exchange format. ACH is a batch electronic funds transfer network with entries such as ARC, BOC, POP, CCD, and PPD. ACH does not carry the cheque image; it carries an electronic debit or credit entry. They can both originate from a cheque workflow, but they are different rails and different data contracts.

Does Chequedb generate X9.37 or ACH files?

Chequedb extracts and validates cheque fields and produces reviewable records. X9.37 generation, ACH origination, transmission, and clearing are deployment- and integration-specific. Confirm the current scope with the team; do not assume the OCR response is a cash letter or an ACH file.

What fields should be reconciled between the image and the file?

At minimum: routing number, account number, cheque number, amount, date, payee, and the item reference. Also reconcile bundle/sequence or trace numbers, return codes, and the deposit/ledger reference. Keep the image, extracted fields, file row, and bank response linked.

What is the biggest integration mistake?

Treating a successful OCR extraction as proof of deposit or settlement. Extraction is an input to review and clearing. The bank or payment operator makes the acceptance, posting, return, and settlement decision. Build the audit trail around that boundary.

This guide is technical education, not legal or financial advice. Confirm clearing, presentment, and settlement rules with the relevant bank or payment operator.