Home › Blog

How to find duplicate files on a Mac without deleting the wrong one

25 August 202610 min read

Somewhere on your Mac there is a file called invoice final FINAL v2 (1).pdf. There are also three other copies of it, and at least one of them is not actually a copy.

Duplicate files are the least dramatic storage problem and the most annoying one to fix, because unlike a 60 GB virtual machine you cannot solve it with a single decision. There are four hundred of them, each one needs a judgement, and roughly twelve of those judgements actually matter.

Worse, a meaningful share of what any duplicate finder shows you are not duplicates, and some of the real ones are not costing you a single byte. Both of those are worth understanding before you start deleting.

The six ways you ended up with four copies

Duplicates are not carelessness. Every one of these is a normal thing happening normally.

The browser did it. Download the same PDF twice and you get statement.pdf and statement (1).pdf. The browser is being careful not to overwrite something you might want. It is right to do this and the result is still clutter.

Cloud sync resolved a conflict. Edit a document on two devices before they sync and Dropbox, iCloud or OneDrive keeps both, with the loser renamed to something like Notes (conflicted copy 2025-03-14).md. This is the correct behaviour and infinitely better than silently discarding your work, but nothing ever cleans it up afterwards.

You imported the same photos twice. Camera cards get imported, then imported again after you were not sure the first one worked.

Someone made a backup folder. Project, then Project old, then Project old 2, made at the moment before attempting something risky. The risky thing worked. The folders remain.

An email attachment was saved more than once. Mail keeps its own copy inside the message store, and you saved another to Downloads, and possibly a third to Desktop when you needed to attach it to something else.

A restore put things back that were already there. Migration Assistant and Time Machine restores are the single largest producer of duplicates, because they faithfully return files to a Mac that in many cases still had them.

Proving two files are actually identical

This is where most duplicate finders are bad, and where the difference between tools is real rather than cosmetic.

Matching by name does not work

Two files called IMG_4471.jpg from two different cameras are different photos. One file called report.pdf and another called report copy.pdf may be identical. Names carry almost no information about content, in either direction.

Matching by size barely works

Same size is necessary but nowhere near sufficient. Every empty file is zero bytes. Every photo from the same camera at the same settings clusters within a narrow band. Uncompressed audio of the same length is the same size regardless of what the audio is. Size is useful only as a first filter to decide what is worth looking at more closely.

Matching by content is the only honest answer

The reliable method reads the bytes and produces a hash, a fixed length fingerprint where identical input always gives identical output. Two files with the same hash are the same file, for any practical purpose.

Doing this naively is slow, because it means reading every byte of every file on the disk. The standard approach is three passes, each cheaper than the one after it:

1. Group by size. Anything with a unique size cannot have a duplicate, and this eliminates the large majority of files without reading any content at all.
2. Within each size group, hash the first few kilobytes. Files that differ near the start, which is most of them, drop out here.
3. Only for the survivors, hash the whole file.

You can do a single comparison yourself:

shasum -a 256 file-one.pdf file-two.pdf

Matching output means identical files. This is fine for checking two things and completely impractical for checking forty thousand, which is the entire reason dedicated tools exist.

The duplicates that cost you nothing

Now the part that catches people out, and the reason a duplicate finder can report "40 GB recoverable" and then recover 3 GB.

APFS supports clones. When you duplicate a file in Finder, macOS frequently does not copy any data. It creates a second name pointing at the same blocks on disk, and only writes new data if one of the two is later modified.

So you have two files. Both report 4 GB. Both are genuinely, byte for byte, identical. Together they occupy 4 GB of disk, not 8. Delete one and you free nothing at all, because the blocks are still in use by the other.

Hard links behave the same way for the same reason. Any tool that adds up the logical sizes of the duplicates it found will confidently overstate what you stand to gain, sometimes by an enormous margin. This is the same distinction between logical and allocated size covered in the article on finding large files, and it matters more here than anywhere else.

Choosing which copy to keep

Most tools default to keeping the oldest, or the newest, or the one with the shortest path. All three are bad defaults, because file dates on a Mac are close to meaningless. Copying a file, syncing it, restoring it from a backup or unzipping it can all rewrite the creation date. The "original" is frequently dated later than the copy.

A better rule: keep the one you would look for first.

A file in ~/Documents/Tax/2025/ is worth more than a byte identical copy in ~/Downloads/, not because it is older but because in eighteen months you will find it. That is the entire value of a file you are keeping for reference.

In descending order of what to keep:

1. Anything in a deliberately organised folder structure.
2. Anything in Documents, Desktop or Pictures.
3. The copy without (1), copy or conflicted in its name.
4. Anything in Downloads, which is a waiting room and not a filing system.

Four places to never run a duplicate finder

This is the section that prevents the bad afternoon. In all four of these, duplicate files exist deliberately and removing them breaks things.

A process that will not hurt you

1. Scan only Documents, Downloads, Desktop and Movies to begin with. Nothing system, nothing application managed.
2. Sort results by recoverable space, not by count. Forty duplicate text files are not worth a single decision; one duplicated video might be.
3. Ignore anything under a megabyte on the first pass. The time you spend deciding is worth more than the space.
4. Send everything to the Trash rather than deleting outright.
5. Leave the Trash for a week. If nothing has gone wrong, empty it.

Step three is the one that saves your evening. There is a strong temptation to clear every duplicate because the number is satisfying to reduce, and the honest maths is that ninety percent of the entries account for two percent of the space.

Common questions