wget Cheatsheet - Download, Resume & Mirror Commands

In scripts, what determines whether a download works is not the transfer itself but what happens after it fails, and wget handles that far better than a browser does. The centrepiece is -c: re-running the same command after an interruption sends a Range header and resumes from the last received byte, which is the difference between success and failure on multi-gigabyte artifacts and mirror syncs. Pair it with --tries and --wait to bound retries, --timeout to avoid hanging on a dead connection, and --retry-on-http-error so recoverable 503s and 429s re-enter the retry loop. The recursive group covers -r depth, -np to stay off parent directories, -A and -R filtering, and -e robots=off. The curl boundary is worth drawing: curl wins for a one-shot fetch whose output should flow into a pipeline, while wget is right when you need resume, recursion, mirroring, or a finished file on disk. Whichever you pick, in unattended jobs the first thing to run out is local disk space, not bandwidth.

SysOps·38 commands·Last updated 2026-07-21

Basic Download 7

wget URL

Download a file to the current directory

wget -O filename URL

Download and save as a given name

wget -P /path/to/dir URL

Download to a given directory

wget -b URL

Download in the background

wget -q URL

Quiet download (no progress)

wget -v URL

Verbose output

wget -nc URL

Skip existing files (no-clobber)

Resume 3

wget -c URL

Resume a partial download

wget -c -O filename URL

Resume and save as a given name

wget --continue URL

Resume (same as -c)

Recursive 8

wget -r URL

Recursive download (default depth 5)

wget -r -l 2 URL

Recursive, depth 2

wget -r -np URL

Recursive, no parent directories

wget -r -nd URL

Recursive, no directory structure

wget -r -N URL

Recursive, newer files only

wget -m URL

Mirror a site (equiv. -r -N -l inf --no-remove-listing)

wget --mirror URL

Mirror a site (same as -m)

wget -r -H -D example.com URL

Cross-host but limit to a domain (-H host, -D domain)

Filter 6

wget -A "*.pdf,*.doc" URL

Download only given types

wget -R "*.tmp,*.log" URL

Reject given types

wget --accept-regex ".*\.pdf$" URL

Accept by regex

wget --reject-regex ".*\.tmp$" URL

Reject by regex

wget -I /dir1,/dir2 URL

Download only given directories

wget -X /dir1,/dir2 URL

Exclude given directories

Auth & Proxy 6

wget --user=username --password=pass URL

HTTP authentication

wget --ftp-user=user --ftp-password=pass URL

FTP authentication

wget -e use_proxy=yes -e http_proxy=URL

Use an HTTP proxy

wget -e use_proxy=yes -e https_proxy=URL

Use an HTTPS proxy

wget --no-check-certificate URL

Skip SSL certificate verification

wget --user-agent="Mozilla/5.0" URL

Custom User-Agent (bypass UA checks)

Common Patterns 8

wget -i urls.txt

Download from a list of URLs

wget -r -l 1 -A "*.jpg,*.png" URL

Download all images on a page

wget -r -l 1 --convert-links URL

Download and rewrite links to local

wget -r -l 1 -k -E -p URL

Download a full page (images, CSS, JS)

wget --limit-rate=100k URL

Limit download speed

wget --timeout=30 --tries=3 URL

Set timeout and retry count

wget --spider URL

Check only, don't download (link test)

wget -w 2 --random-wait -r URL

Polite crawl: 2s wait plus random jitter

Command Examples

Resume a large download

wget -c https://example.com/large.iso

-c makes wget send a Range header and pick up where it stopped, and it behaves far better next to --timeout=30 --tries=3. The filename has to stay the same one as before, otherwise -c finds nothing to resume and starts over.

Output

Resuming last request: Length: 1048576000 [1048576000/1048576000]
HTTP request sent, awaiting response... 206 Partial Content
Length: 1048576000 [1039460352/1048576000]  30.0MB/s  eta 00:00:03
1048576000  100%[====] 103.95MB  29.87MB/s

Mirror a page including images and CSS

wget -r -l 1 -k -E -p https://example.com/docs/

-k rewrites the page links to local paths, -E saves the pages as .html, -p pulls in images and other assets, and -l 1 caps the recursion at a single level so the crawl cannot run away.

Output

--2026-09-30 15:20:11--  https://example.com/docs/
Resolving example.com (example.com)... 93.184.216.34
Connecting to example.com|93.184.216.34|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 12480 [text/html]
Saving to: ‘example.com/docs/index.html’

2026-09-30 15:20:12 (18.0 MB/s) - 12480 [text/html] saved

Download by file type with a rate limit

wget -r -l 2 -A "*.pdf,*.zip" --limit-rate=500k -P ~/downloads https://example.com/files/

-A is the allow list and only the listed types are kept, while --limit-rate is expressed in KB/s, so 500k throttles to roughly 500KB/s. Add -np as well, otherwise the crawler follows parent links and walks straight off the site.

Common Pitfalls

wget -r without a depth limit or -np can walk off the site and try to download a large part of the internet.
--no-check-certificate disables verification, allowing man-in-the-middle replacement of downloaded content; debug-only.
-c resume requires the same filename as the original attempt — renaming restarts the download, and it silently does nothing if the server ignores Range.
Probe recursive downloads with -l 1 first and confirm the directory structure before widening the scope.
Check robots.txt and the site terms before recursive downloading; bulk crawling may violate them.

Tips

wget supports HTTP, HTTPS and FTP, with resume and recursive download.
Mind the depth (-l) on recursive downloads to avoid pulling too many files.
Use -m to mirror a site; it auto-handles link conversion and timestamps.

Official References

Each command links to its official documentation below, so you can verify the latest usage and read deeper.

Linux man-pages

Maintained by LaoHand

Publicly updated on Jul 21, 2026, continuously proofread against official docs.

Contact Us

Wrong command or description? Send us corrections, business inquiries or product feedback by email.