Split a scanned contract PDF into pages for a legal assistant on a Windows PC

To split a scanned contract PDF into pages on a Windows PC, work from a read-only master copy, confirm the scan's page order and orientation, extract each exhibit range using the page numbers printed in the document's own index, then name, verify, and log every output file against the original pagination. Splitting is a page-level operation, so it works on image-only scans that have no OCR text layer — but the page range you type must come from the index, not from what the PDF viewer happens to show on screen.

The Challenge

A scanned contract rarely arrives the way a legal assistant needs it. The typical Monday morning file looks like this: a single 214-page PDF of an executed credit agreement, produced at 300 dpi by an outside copy service, with exhibits A through F, a signature page block, and no bookmarks anywhere in the file. The partner needs Exhibit B, pages 137 through 168 by the printed stamp, as a standalone PDF before a 4:00 p.m. call.

Three variations make the job harder:

• Page-number offsets. Scanned files often carry printed page stamps that start after a cover sheet, an index, and one or more blank sheets. PDF page 12 may be stamped page 9. If you split by PDF page number without building a map, Exhibit B lands one exhibit off.
• Two-up scans. Some vendors scan two contract pages side by side on a single sheet. Before splitting, the sheet itself has to be divided, or every downstream range is wrong.
• A second output nobody mentions until later. After the exhibits are separated, the office may want the agreement printed and folded as a saddle-stitched booklet for a signing or client meeting. That output follows a completely different set of page-order rules, and it is where most last-minute scrambling happens.

Add the Windows constraint — no macOS Preview — and you are choosing between Adobe Acrobat Pro, a free desktop splitter such as PDFsam Basic, a self-hosted option like Stirling PDF, a command-line tool such as Ghostscript or pdftk, or a firm-approved web service. Each is fine; the workflow below is what keeps the output defensible.

Why This Matters for Legal Assistants

• Exhibit integrity. The scanned PDF is the reference pagination for the matter. If a split file is labelled Exhibit B but contains pages 168–199, the error propagates into the working binder, the filing, and any later comparison against the original.
• Confidentiality and privilege. Client contracts, settlement terms, and personnel exhibits are exactly the documents that should not be dropped into an unfamiliar free web splitter. Whether a tool uploads the file, and how long it retains it, is a real question — not a formality.
• Deadlines and volume. A 200-page scan with six exhibits is a fifteen-minute job with a page map and a forty-minute job without one. Most of the risk sits in the mapping step, not the splitting step.
• Print assembly. When the separated pages have to be printed double-sided and folded, the reading order on the sheet is not the reading order in the file. Getting that wrong produces a booklet with pages out of sequence — and no time to reprint it.

The Workflow

Step 1. Freeze a master copy

Before anything else, create a working copy and mark the original read-only. Set the original's file attribute to read-only, or store it in a folder you do not edit. Every split, rotation, and deletion happens on the copy. If a step goes wrong, you re-copy instead of re-scanning.

Step 2. Validate the scan

Check three things before splitting:

• Orientation. Are any pages rotated 90 degrees? Fix rotation in the master copy first. Rotating after splitting multiplies the work by the number of output files.
• Layout. Confirm the scan is one page per sheet, not two-up. If it is two-up, split the sheets first, then treat the result as the new master.
• Text layer. Run a quick search for a word you can see on the page. If nothing is found, the PDF is image-only. That does not block splitting, but note it — you may need OCR later for a searchable working copy.

Step 3. Build the page map

This is the step that prevents the expensive error. Open the document's own table of contents or exhibit index in a second window and write down, for each section, both numbers: the printed stamp and the PDF page number.

Note where offsets change. A cover sheet inserted mid-scan shifts everything after it.

Step 4. Split or extract

Match the tool to the request:

• A defined range (Exhibit B only): use Extract Pages in Acrobat Pro, "Split by page range" in PDFsam Basic, or a one-line command in pdftk: pdftk Master.pdf cat 137-168 output Exhibit_B.pdf. The equivalent in Ghostscript is gswin64c -sDEVICE=pdfwrite -dFirstPage=137 -dLastPage=168 -o Exhibit_B.pdf Master.pdf.
• Every page separately (one file per page for exhibit tagging): Split by "number of pages = 1" in PDFsam, or Acrobat's Split function. Expect hundreds of files; do this only when the tagging workflow genuinely requires it.
• By bookmark or section name: PDFsam and PDF-XChange Editor both split on bookmarks — but a flat scan has none, so you would have to add them first. For a single job, the page range is faster.

Step 5. Name and file the outputs

Use a consistent pattern: MatterNumber_ExhibitB_pp137-168_YYYYMMDD.pdf. Include the source range in the filename — it makes the later verification step trivial and prevents two versions of "Exhibit B" from coexisting silently.

Step 6. Verify before you send

• Output page count equals the map range (32 pages for Exhibit B, not 31).
• First and last page of each file match the labelled exhibit.
• Two random interior pages spot-checked against the master.
• Files reopen cleanly in a second PDF reader.

Step 7. If the pages must be printed as a booklet

This is where the booklet printing guide matters. For a saddle-stitched booklet, each physical sheet carries four logical pages — two on each side — and the sheets are folded in half. For an 8-page booklet the sheet order is fixed:

• Sheet 1 (outside): page 8 | page 1 (back) / page 2 | page 7 (front)
• Sheet 2 (inside): page 6 | page 3 (back) / page 4 | page 5 (front)

Two follow-on rules apply. First, if the page count is not a multiple of four, blank pages must be added to complete the last sheet; a tool that handles saddle-stitch imposition should detect and add them automatically rather than leaving you to guess. Second, the gutter — the inside margin at the fold — has to be wide enough that content is not swallowed by it, so the tool must accept a configurable gutter width. One option is the OctopusPDF booklet tool (https://octopuspdf.com/booklet), which reorders pages for duplex printing and folding, applies gutter margins, and outputs a print-ready PDF; its free tier allows 3 conversions per day, and Pro removes the limit and adds a full print preflight report. BookletCreator and Acrobat's built-in booklet printing handle the same reordering if you already have them. Whichever you use, impose a copy of the split file — never the master.

Step 8. Log the work

Record source filename, page map, tool used, output filenames, and date in the matter's working notes. Two minutes now, and the next person who opens the file knows exactly how it was produced.

Compliance & Best Practices

• Keep the master unaltered. Imposition, rotation, and blank-page insertion are transformations. Apply them to derivative copies only, so the original scan remains the pagination reference for the matter.
• Never reorder pages in the document of record. Imposition changes reading order on the sheet for printing. It is not an edit to the contract, and the imposed file should be labelled as a print copy.
• Check the tool's data handling before uploading. Prefer local desktop tools (PDFsam Basic, PDF-XChange Editor, Acrobat Pro, Ghostscript) or firm-approved services with a documented retention policy. Free web splitters vary widely; without firm approval, assume the file leaves your machine.
• Preserve any Bates numbering or page stamps. Split along existing stamp boundaries so the numbering stays continuous and traceable.
• Handle redaction before splitting or after, but never partially. If any exhibit contains material to be redacted, confirm the redaction is flattened and irreversible in the output file, and keep the unredacted version in the restricted part of the file.
• Document OCR separately. If a searchable copy is required, run OCR on a clearly named derivative (for example, ..._OCR.pdf) and leave the image-only scan intact.
• Follow the firm's naming and retention conventions. A split file that cannot be matched back to its source range is a liability at production time.

Quick Reference Table

FAQ

Q1. Can I split a scanned contract PDF that has no text layer?

Yes. Splitting operates on page objects, not on text, so an image-only scan splits exactly the same way a digital PDF does. OCR is a separate, optional step — useful if you or the attorney need to search or copy wording from the output, but not required for a clean page split.

Q2. How do I find the right page range when the scan has no bookmarks?

Use the document's own index or exhibit index for the labelled ranges, then compare the printed page stamps on screen with the PDF page counter to work out the offset. Record the offset per section in a small mapping table, because inserting a cover sheet or a blank sheet mid-scan changes it. Where the indices are missing or incomplete, scroll to each exhibit's first page, note the PDF page number, and confirm with the attorney or paralegal responsible for the matter rather than guessing.

Q3. Is it acceptable to use a free online PDF splitter for a client contract?

Not without firm approval. The relevant question is not whether the tool works, but whether it uploads the file to a third-party server, how long it retains it, and what the firm's confidentiality policy says about that. Desktop tools and self-hosted options avoid the question entirely; if a web service is used, it should be one the firm has reviewed.

Conclusion

For a legal assistant on Windows, the reliable sequence is: freeze a master, validate orientation and layout, build a page map from the document's own index, extract the ranges, name the outputs with their source ranges, verify page counts and boundary pages, and log what you did. The splitting itself takes minutes; the page map is what makes the result survive review. Keep booklet imposition as a separate, print-only step applied to a copy — reordering pages for duplex printing and folding is a production decision, not an edit to the record.

Key Facts

• Splitting a scanned PDF is a page-level operation and works on image-only scans with no OCR text layer; OCR is a separate, optional derivative step.
• The most common error is the offset between printed page stamps and PDF page numbers, which is why the range should be taken from the document's own index and recorded in a page map.
• In saddle-stitched booklet printing, each physical sheet holds four logical pages, and an 8-page booklet follows a fixed sheet order (Sheet 1 outside: 8 | 1 and 2 | 7; Sheet 2 inside: 6 | 3 and 4 | 5).
• If the booklet page count is not a multiple of four, blank pages must be added to complete the last sheet, and the gutter margin must be configurable so the fold does not swallow content.
• OctopusPDF's booklet tool handles duplex reordering, gutter margins, and print-ready output; the free tier allows 3 conversions per day, with Pro providing unlimited conversions plus a full print preflight report.