No, not every output part is guaranteed to sit below the target when you split a PDF by size — multi-page parts are measured and capped, but a single page that already exceeds the target on its own is kept as a flagged one-page part rather than being silently dropped or infinitely retried. The guarantee covers any group of consecutive pages that the tool can build, serialize, and measure: as soon as the next page would push the candidate beyond the MiB target you entered, the current group is finalized and a new part begins. For a single page that cannot fit, the tool treats it as a known exception: it ships that page alone, marks the download as oversized, and stops. This exception is what prevents silent page loss and prevents the planner from looping forever trying to fit an unplaceable page. If a system downstream rejects that flagged file, the remedy is to raise the target, split differently, or compress the source first — not to assume the tool will eventually squeeze a too-large page into compliance. The behavior is deliberate, local to your browser, and the same on every run for the same input and target.
What the Size Guarantee Actually Covers
The size guarantee applies to anything the planner can build as a multi-page candidate. The tool copies consecutive pages into a candidate PDF, serializes that candidate to count its real bytes, and accepts the candidate only when its measured length is at or below the target you set. If adding the next page would push the candidate past the target, the current candidate is finalized as a part and a new candidate starts with the next page. Every part produced this way has a measured byte length at or below the target at the time it was built. The page counts of all accepted parts always add up to the page count of the source file because pages are never reordered, duplicated, or skipped.
| Scenario | Is the resulting part below the target? | Why |
|---|---|---|
| Group of consecutive pages that fit | Yes | The serialized candidate is measured and accepted only when its bytes are at or below the target. |
| The last page pushes the candidate over the limit | Yes for the prior group | The candidate is finalized before the overflow page is added; a new part begins with the overflow page. |
| One page is already larger than the target on its own | No — and flagged | The page is kept alone in a clearly labeled one-page part; this is the only documented exception. |
| Target changed mid-run | Yes, against the new target | Prior download links are revoked; the planner reruns against the current target and a job identifier prevents a slow earlier result from replacing the new one. |
When a Part Can Exceed Your Target
The single documented exception is the unplaceable page. PDF resources can be shared across pages, compressed differently, or copied into a new document with a different overhead — so the planner measures every candidate rather than assuming each page consumes an equal slice. Sometimes the measurement returns a number the planner cannot reduce: a single page that already serializes to more bytes than the target you typed. In that case the page goes into its own one-page part, the download is visibly labeled as exceeding the target, and the planner moves on. This is the only time you will see a part that is larger than the target you set, and the labeling is intentional so the file cannot be mistaken for a successful fit.
If that flagged file is a problem for a downstream system — an email gateway with a 25 MB cap, an upload portal, a form with a hard limit — the right next step depends on what you actually control. Raise the target so the flagged page fits, then re-run the split. Run a dedicated compression step on the source first, then split the compressed file with a smaller target. Use a page-based splitter such as Split PDF in Half or Split PDF when you care about balanced page counts rather than byte caps, or Extract PDF Pages if the issue is concentrated in a small number of heavy pages. The flagged exception is what prevents two worse outcomes: a page that disappears from the output without warning, and a planner that retries the same oversize page forever.
How to Split a PDF by Size in Your Browser
The workflow is short because everything — page copying, serialization, measurement, and download creation — runs locally in the current tab. No part of the document leaves your device.
- Choose a non-empty PDF no larger than 25 MiB. Empty, encrypted, damaged, or oversized files produce a visible error instead of a partial result.
- Enter the maximum generated size per part from 0.01 to 25 MiB. Decimal input is accepted to three places; the minimum is useful for deterministic testing and very small documents.
- Select Split PDF by Size. The tool copies pages into candidate PDFs, serializes each candidate to obtain its real byte length, and starts a new one before the candidate crosses the target.
- Review any oversized single-page warning. If a page cannot fit on its own, it is shipped alone in a flagged one-page part rather than being dropped.
- Download each numbered part. The tool also displays each generated part's byte length so you can verify the result before relying on the file.
Why the Tool Measures Generated Bytes Instead of Estimating
A naive splitter divides the source file size by the number of pages and cuts at fixed page counts. That estimate ignores how PDFs actually behave. Copied pages can carry shared font subsets, reused image XObjects, cross-reference streams, and trailer overhead that the original file already paid for once. When those pages are copied into a new document, the serializer may inline resources it deduplicated before, may recompress images at a different quality, and may write a fresh cross-reference table whose size depends on the number of objects in the new file. The result is that two pages of identical-looking content can serialize to noticeably different byte counts in different output documents.
By serializing each candidate and measuring its real length, the planner catches these effects honestly. The reported byte length is the actual length of the file the planner would write to disk, not a proportional guess. That is also why the browser and operating-system download displays can round the same byte count differently, and why the tool displays each part's size next to its download so you have an independent reference number. For more on the measurement approach and what it means in practice, see Split a PDF by File Size Using Measured Output Bytes.
Limits, Inputs, and Files the Tool Will Reject
Two interacting limits decide whether the tool can complete the job. The input must be a non-empty PDF no larger than 25 MiB. The target must be between 0.01 and 25 MiB. If either side is outside its range, no output is produced. The target uses mebibytes — one MiB equals 1,048,576 bytes — so a 2 MiB target is exactly 2 × 1,048,576 = 2,097,152 bytes. That distinction matters when an upload cap is stated in megabytes (1,000,000 bytes) rather than mebibytes: the same number describes a slightly smaller file in MB than in MiB.
The tool also produces visible errors rather than silent failures for empty files, files larger than 25 MiB, encrypted files, damaged files, mislabeled files, and any format it does not support. Password-locked PDFs are not bypassed; the password is not guessed, stripped, or removed. Changing the file or the size target revokes prior download URLs so an older result cannot be confused with the current settings, and a job identifier prevents a slower earlier operation from replacing a newer result.
If your real goal is to fit under a hard upload cap, remember that page-boundary splitting with measured bytes is the only guarantee this tool makes. It does not compress images, optimize fonts, linearize a PDF, or force every output under the target when a single page is already too large. For JPG-heavy inputs that need a specific byte count, run Compress PDF to Size first; for general reduction without a target, Compress PDF is the local alternative. For equal page counts use Split PDF in Half, for manual ranges use Extract PDF Pages, and for layout adjustments without changing bytes use Resize PDF.
For a deeper look, see Plan the Steps Needed to Compress a PDF to Size.