The best duplicate file finder for every cleanup job

1. dupeGuru — best for most people

dupeGuru gets the fundamentals right: content-based matching, a fuzzy filename mode for "almost the same name" cases, dedicated music and picture modes, and — most importantly — safe defaults. It marks one file per group as the reference you can't accidentally delete, and sends removals to the trash. It's free, open source, and runs on Windows, Mac and Linux. In our testing it cleared a deliberately messy 200 GB folder set without a single false positive in standard mode. The honest cons: development has slowed in recent years, the interface feels dated, and Mac builds sometimes lag behind OS releases. None of that changes the recommendation for a first cleanup.

2. Duplicate Cleaner Pro — best selection tools and photo mode

Finding duplicates is the easy half; choosing which copy to keep across 3,000 groups is where cleanups stall. Duplicate Cleaner Pro (from DigitalVolcano, ≈$40 one-time) has the best answer we've used: a selection assistant that marks files by rule — keep newest, keep the one in this folder tree, keep highest-resolution image — so a giant cleanup becomes a few decisions instead of thousands. Its image mode catches resized and rotated copies, and there's an audio mode for music libraries. The free version is genuinely useful but limited (notably around image matching), and it's Windows-only. If you're untangling a photo library measured in years, this is the one paid tool on the page that earns its price.

3. AllDup — most configurable free option

AllDup is free (donationware) and lets you define "duplicate" almost any way you want: any combination of content hash, size, name, extension and dates, plus similar-image and similar-audio matching. Its standout power feature is replacing duplicates with hardlinks — the file appears in both places but occupies disk space once — which is the right move when two programs each insist on their own copy. The cost of all that power is a dense, old-fashioned interface with enough options to be genuinely intimidating; check every setting before your first run, because flexibility cuts both ways. Windows-only.

4. Czkawka — fastest scans, open source

Czkawka (Polish for "hiccup") is a modern open-source tool written in Rust, and it's the fastest scanner we've tested — a full pass over our 1.2-million-file library finished in minutes where older tools took most of an hour. It uses the textbook staged pipeline (size, then partial hash, then full hash), finds similar images and broken files, and also reports empty folders and largest files, overlapping nicely with our large-files guide. The trade-offs: documentation is sparse, the interface assumes some confidence, and it has fewer guardrails than dupeGuru — there's no strong concept of a protected reference copy, so the safety rules below matter even more here.

5. Our File Finder — quickest single-folder check

If you suspect one specific folder — Downloads, usually — you don't need to install anything. Our free in-browser File Finder has a duplicate mode that hashes file contents right in your browser, with nothing uploaded anywhere. To be clear about its limits: it works on a single folder tree at a time and doesn't do similar-image matching, so it won't replace the desktop tools for a whole-drive cleanup. But for "is this folder full of repeats?" it's the fastest honest answer on this page — try it on Downloads first.

We seeded our 1.2-million-file test library with controlled duplicate sets: exact copies with different names and dates, resized and re-saved versions of the same photos, RAW+JPEG pairs, and near-miss files that differ by a single byte. Each tool ran the same scans on the same Ryzen 7 / 32 GB / NVMe machine (full spec on our about page), and we scored three things: did it find every planted exact duplicate, did it flag any near-miss as a false positive, and how hard the tool made it to do something destructive. All four desktop tools found 100% of the exact duplicates — staged hashing is mature technology — so the rankings above turn on speed, image matching and selection workflow rather than raw accuracy. The single-byte near-misses fooled nobody in exact mode, which is exactly why we trust content hashing and not filenames.

• Never auto-delete. Every tool here has some one-click clean option. Don't use it. Review what's selected in every group — rules make mistakes, and so do similarity algorithms.
• Recycle Bin first, always. Configure deletions to go to the Recycle Bin (or a quarantine folder), live with the cleanup for a week, then empty it.
• Keep one copy by definition. Use the tool's reference/protection feature so at least one file per group can't be selected. dupeGuru does this by default; in Czkawka you enforce it yourself.
• Exclude system and app folders. Windows, Program Files, AppData and application libraries (Lightroom catalogs, music app databases) contain intentional duplicates. Deduping them breaks things.
• Mind cloud-sync folders. In OneDrive, Dropbox or Google Drive folders, deletions sync everywhere and online-only placeholders may re-download just to be hashed. Pause syncing before you scan.
• Back up before a big cleanup. If you're about to delete tens of gigabytes, copy the target folders to an external drive first — our external drive picks are cheap insurance against an irreversible mistake.

A note on photos, the most common duplicate problem: exact-hash matching misses resized exports, edited re-saves and RAW+JPEG pairs, so use an image mode (dupeGuru picture mode, Duplicate Cleaner Pro, Czkawka) with a strict similarity threshold first, then loosen it gradually. Treat burst shots carefully — visually "similar" is not "duplicate" when one frame is the keeper. And once the cleanup is done, a sane folder structure is what stops the duplicates coming back; pairing it with a fast search tool from our free tools roundup means you find files instead of re-copying them.