Find the Largest Files on Your Mac (Simple Script & Tools)

Running out of disk space on your Mac but can’t figure out where it all went? You’re not alone. Apple’s own storage tools in macOS Sequoia and Tahoe give you a high-level overview but often hide the real culprits: orphaned AI models, Time Machine snapshots, and APFS clones that report misleading sizes. Third-party “disk cleaner” apps charge subscriptions for what a few terminal commands can do for free.

This guide covers five ways to find large files on Mac: a zero-setup Finder search, the instant mdfind Spotlight CLI, an improved Bash script for full system scans, modern interactive CLI tools, and the hidden space eaters that most guides ignore. Everything here works on macOS Sequoia 15.x and Tahoe (macOS 26).

If you’re looking for free Mac apps beyond disk cleanup, check toolhunt.net mac apps section.

Quick method: Finder search (no Terminal needed)

The simplest way to find large files uses Finder’s built-in search. No Terminal, no scripts, no installs. This is enough for most casual cleanup.

Finder search has limits: it won’t show hidden directories (~/Library, ~/.ollama), system files, or files inside app bundles. For those, you need Terminal methods.

Spotlight CLI: mdfind for instant results

mdfind queries the Spotlight index, making it near-instant compared to walking the filesystem with find. It’s the fastest way to locate large files if Spotlight indexing is enabled.

# Find all files > 1 GB using Spotlight index (near-instant)
mdfind "kMDItemFSSize > 1000000000" -onlyin ~/ 2>/dev/null | \
  while IFS= read -r f; do
    stat -f '%z %N' "$f" 2>/dev/null
  done | sort -rn | head -20

# Find all files > 500 MB in your home directory
mdfind "kMDItemFSSize > 500000000" -onlyin ~/ 2>/dev/null | \
  while IFS= read -r f; do
    stat -f '%z %N' "$f" 2>/dev/null
  done | sort -rn | head -20

The output shows byte sizes and full paths, sorted largest-first. You can pipe it through numfmt --to=iec for human-readable sizes, though stat on macOS doesn’t have the GNU numfmt by default.

Caveat: mdfind only finds files that Spotlight has indexed. Some system directories and external drives may not be indexed. For a comprehensive scan, use the Bash script below.

The Bash script: full system scan

This is the core method — an improved macOS-optimized script that scans your filesystem, reports disk usage, and lists the largest files. The original version of this script had issues with modern macOS permissions and scanned external drives. This version fixes both.

Prerequisites: Full Disk Access

Since macOS Mojave (10.14), even sudo can’t access certain user directories without Terminal having Full Disk Access. Without it, the script silently skips entire directory trees. The 2>/dev/null hides the “Operation not permitted” errors.

Grant Full Disk Access:

1. Open System Settings > Privacy & Security > Full Disk Access
2. Click the + button
3. Navigate to /Applications/Utilities/Terminal.app (or your terminal of choice, like Ghostty or WezTerm)
4. Toggle it ON
5. Quit and reopen the terminal for changes to take effect

The improved script

Key changes from the original:

• Scans your home directory by default (where most user files live), with --all flag for power users
• Uses -x to stay on one filesystem (won’t wander into external drives or network mounts)
• Excludes noisy directories that waste scan time
• Safer output parsing with IFS=$'\t'
• Adds a Time Machine snapshot check at the end

#!/bin/bash

# Find the largest files on your Mac
# Usage: ./find_large_files.sh [number_of_files]
#        ./find_large_files.sh --all [number_of_files]  (scan entire system)

# Parse arguments
scan_path="$HOME"
num_files=20

if [[ "$1" == "--all" ]]; then
    scan_path="/"
    shift
fi

if [[ -n "$1" && "$1" =~ ^[0-9]+$ ]]; then
    num_files=$1
fi

# Print disk space information
echo "==========================================="
echo "DISK SPACE INFORMATION"
echo "==========================================="
df -h "$scan_path" | awk 'NR==2 {
    printf "Total Space: %s\n", $2
    printf "Used Space:  %s\n", $3
    printf "Free Space:  %s\n", $4
    printf "Usage:       %s\n", $5
}'
echo "==========================================="
echo

# Build exclusion list. Skip directories that waste time or produce noise
excludes=(
    -not \( -path "/System/*" -prune \)
    -not \( -path "/Volumes/*" -prune \)
    -not \( -path "/private/var/vm/*" -prune \)
    -not \( -path "/private/var/folders/*" -prune \)
    -not \( -path "*/Library/Caches/*" -prune \)
    -not \( -path "*/Library/Containers/*" -prune \)
    -not \( -path "*/node_modules/*" -prune \)
)

echo "Searching for the $num_files largest files..."
echo "Scanning: $scan_path"
echo "This may take a minute on large drives."
echo

# Create temporary file for results
tmp_file=$(mktemp)

# Run the scan. -x keeps us on one filesystem
# sudo is needed when scanning / or other users' directories
if [[ "$scan_path" == "/" ]]; then
    sudo find "$scan_path" -x "${excludes[@]}" \
        -type f -print0 2>/dev/null | \
        xargs -0 du -h 2>/dev/null | \
        sort -rh | \
        head -n "$num_files" > "$tmp_file"
else
    find "$scan_path" -x "${excludes[@]}" \
        -type f -print0 2>/dev/null | \
        xargs -0 du -h 2>/dev/null | \
        sort -rh | \
        head -n "$num_files" > "$tmp_file"
fi

# Print formatted results
echo "==========================================="
echo "TOP $num_files LARGEST FILES"
echo "==========================================="
echo
printf "%-8s | %s\n" "Size" "File Path"
echo "-------------------------------------------"
while IFS=$'\t' read -r size file; do
    printf "%-8s | %s\n" "$size" "$file"
done < "$tmp_file"
echo "==========================================="

rm "$tmp_file"

# Check for Time Machine local snapshots
echo
echo "==========================================="
echo "TIME MACHINE LOCAL SNAPSHOTS"
echo "==========================================="
snapshots=$(tmutil listlocalsnapshots / 2>/dev/null)
if [[ -n "$snapshots" ]]; then
    count=$(echo "$snapshots" | wc -l | tr -d ' ')
    echo "Found $count local snapshot(s). These are invisible to find/du:"
    echo "$snapshots" | head -5
    if (( count > 5 )); then
        echo "  ... and $((count - 5)) more"
    fi
    echo
    echo "To reclaim this space: sudo tmutil thinlocalsnapshots / 999999999999 4"
else
    echo "No local snapshots found."
fi
echo "==========================================="

echo
echo "Scan complete!"

Run the script step-by-step

1. Open Terminal. Find it in Applications > Utilities or press Cmd + Space and type “Terminal”. If you want a better terminal experience, consider setting up Fish shell on macOS for better autocomplete and syntax highlighting.

2. Create the script file:

   nano ~/find_large_files.sh

   Paste the script above, then save with Ctrl + X, Y, Enter.

3. Make it executable:

   chmod +x ~/find_large_files.sh

4. Run it:

# Scan your home directory (default, no sudo needed)
./find_large_files.sh

# Show top 30 files instead of 20
./find_large_files.sh 30

# Scan the entire system (requires sudo for /System, /private, etc.)
./find_large_files.sh --all

# Full scan, top 50 files
./find_large_files.sh --all 50

When running a full system scan, you’ll be prompted for your password. The scan takes 1-5 minutes depending on drive size and whether you have external drives connected.

For long scans, consider running it inside a tmux terminal multiplexer so you can detach and come back later.

Sample output (home directory scan):

===========================================
DISK SPACE INFORMATION
===========================================
Total Space: 460Gi
Used Space:  95Gi
Free Space:  341Gi
Usage:       22%
===========================================

Searching for the 20 largest files...
Scanning: /Users/dragos
This may take a minute on large drives.

===========================================
TOP 20 LARGEST FILES
===========================================

Size     | File Path
-------------------------------------------
13G      | /Users/dragos/.diffusionbee/downloaded_assets/FLUX.1-schnell_flux_schnell_q5p_NNC_all.sqlite
13G      | /Users/dragos/.diffusionbee/downloaded_assets/FLUX.1-dev_flux_dev_q5p_NNC_all.sqlite
8.4G     | /Users/dragos/.ollama/models/blobs/sha256-6e41c39f4490a9e8b7a65916425c6ed97f04ed95bab991c4ab6a462ff84d1608
1.9G     | /Users/dragos/.ollama/models/blobs/sha256-dde5aa3fc5ffc17176b5e8bdc82f587b24b2678c6c66101bf7da77af9f7ccdff
949M     | /Applications/DaVinci Resolve/DaVinci Resolve.app/Contents/MacOS/Resolve
...

===========================================
TIME MACHINE LOCAL SNAPSHOTS
===========================================
Found 3 local snapshot(s). These are invisible to find/du:
com.apple.TimeMachine.2025-01-09-183042.local
com.apple.TimeMachine.2025-01-09-193042.local
com.apple.TimeMachine.2025-01-09-203042.local

To reclaim this space: sudo tmutil thinlocalsnapshots / 999999999999 4
===========================================

If the output shows mostly /System or /Volumes paths, the -x flag or exclusions aren’t working. Check that you’re running the updated script, not the old version.

Verify: The disk info header should show your actual drive size and usage. The file list should contain user-level paths (home directory, Applications), not system internals.

Modern CLI tools: dust, gdu & ncdu

If you find yourself running disk cleanup regularly, a one-shot script isn’t the best tool. These three CLI utilities give you interactive, browsable views of disk usage, much better for exploring and deleting on the fly.

All three install via Homebrew:

brew install dust

dust ~/              # full home directory tree
dust -F ~/           # files only, no directory summaries
dust -n 30 ~/        # show top 30 entries
dust -d 2 ~/         # limit depth to 2 levels (quick overview)

brew install gdu

gdu ~/               # interactive mode. Arrow keys to navigate, 'd' to delete
gdu -t 20 ~/         # non-interactive: top 20 largest items

brew install ncdu

ncdu ~/              # scan and browse interactively
ncdu -x /            # scan root filesystem, don't cross mount points

If you spend a lot of time in the terminal, you can supercharge your Fish shell with plugins for better tab completion on these commands.

Where is my “missing” disk space?

You ran df and it says 200 GB used. You ran du on your home directory and it only adds up to 120 GB. Where did the other 80 GB go? On APFS (the default filesystem since High Sierra), the answer is usually one of three things.

Time Machine local snapshots

Time Machine stores hourly local snapshots directly on your startup disk. These can consume tens of GB but are completely invisible to find, du, and Finder. They don’t show up as files because they’re filesystem-level snapshots.

# Check if snapshots are eating your space
tmutil listlocalsnapshots /

# If you see many snapshots and need space now:
sudo tmutil thinlocalsnapshots / 999999999999 4

The thinlocalsnapshots command tells macOS to purge local snapshots, freeing space immediately. The priority parameter (4) means “low priority.” It won’t interrupt other operations. These snapshots auto-delete when disk pressure occurs, but manual thinning is safe if you need space now.

APFS clones, sparse files & purgeable space

APFS has features that make du output misleading:

• Clones: cp -c (and many apps like pnpm, uv, git worktrees) creates copy-on-write clones. du reports the full size of every clone, but deleting one clone frees zero blocks if another clone exists. A node_modules tree can report 20 GB in du but deleting it frees 3 MB.

• Sparse files: Docker’s Docker.raw disk image and VM images (UTM, Parallels) are sparse files. They report as 64 GB but only occupy the actually-written blocks (e.g., 9 GB). If you’re running Docker, see how to clean up Docker images and reclaim disk space.

• Purgeable space: macOS marks some files as purgeable (caches, local snapshots, iCloud-offloaded files). They count as “used” in du but macOS will auto-delete them when space is needed. You can check purgeable space with:

  diskutil info / | grep -i purgeable

If you’re generating locally AI images like Flux models, those model files can be 10-13 GB each — and DiffusionBee stores them in ~/.diffusionbee/downloaded_assets/.

What’s safe to delete (and what never to touch)

The script finds large files but doesn’t tell you what’s safe to remove. Here’s a practical guide organized by category.

ollama list              # see what's installed
ollama rm <model-name>   # remove a specific model

docker system df         # see Docker's disk usage breakdown
docker system prune -a   # remove all unused images, containers, networks (careful!)
docker image prune       # remove dangling images only (safer)

# Homebrew cache (old downloads)
brew cleanup -s

# npm cache
npm cache clean --force

# pip cache
pip cache purge

# yarn cache
yarn cache clean

# Old iOS/tvOS/watchOS device support (can be many GB)
rm -rf ~/Library/Developer/Xcode/iOS\ DeviceSupport/*

# Old Xcode caches
rm -rf ~/Library/Developer/Xcode/DerivedData/*

# Old simulator data
xcrun simctl delete unavailable

If you’re unsure whether two directories contain the same content before deleting one, you can compare folder contents in Terminal to check.

Related terminal tools

If you’re spending time in Terminal for disk cleanup and system management, these tools make the experience better:

• Ghostty — modern, GPU-accelerated terminal emulator for macOS
• WezTerm — terminal with built-in tmux-like multiplexing
• tmux — terminal multiplexer for long-running scans and multiple sessions
• Fish shell — better default shell with autosuggestions and syntax highlighting
• cmux — AI-assisted terminal sessions for power users

Conclusion

You don’t need paid apps to find and manage large files on your Mac. Between Finder search for quick checks, mdfind for instant results, the Bash script for comprehensive scans, and dust/gdu/ncdu for interactive exploration, you have everything you need.

The real disk hogs are usually AI models (Ollama, DiffusionBee, Stable Diffusion), Docker images, Time Machine local snapshots, and dev project caches. Run the script once a month or whenever you see the “Your disk is almost full” warning. Apple charges $0.99/month for 50 GB of iCloud storage — but cleaning up local files is free.

FAQ

# List snapshots
tmutil listlocalsnapshots /

# Thin them (reclaim space)
sudo tmutil thinlocalsnapshots / 999999999999 4