How to Copy and Rename a File in Linux

Copy a file under a new name

The destination argument to cp is also the new filename, so cp report.txt report-2026.txt copies report.txt into a file called report-2026.txt in one step, with no separate rename command involved. The same syntax works across directories: cp report.txt ~/backups/report-2026.txt copies, relocates, and renames at once. If report-2026.txt already exists, GNU coreutils documents that cp overwrites it without asking, which the next section covers as a real risk rather than a footnote.

Can I copy and rename a file in one command?Yes. cp source newname does both, since the copy destination and the new filename are the same argument.

Copy or rename: what changes on disk

Copying with cp and renaming with mv look similar on the command line but change different things underneath. On the same filesystem, mv only updates a directory entry, matching the behavior documented for the underlying rename system call that mv relies on; cp allocates a new inode and duplicates the file’s data.

For a multi-gigabyte video file already on the same disk, that difference decides your wait time: mv finishes immediately because nothing is copied, while cp has to write the same volume of data a second time before it finishes.

GNU coreutils 9.11, released in April 2026, added an –exchange flag to mv that atomically swaps two files’ names, useful in scripts that update a live file without a moment where neither name exists.

Does renaming a file with mv count as making a copy?No. mv relabels the existing inode on the same filesystem, so no extra disk space is used and any hard links to the file keep working against the same data.

Renaming files safely with mv

mv original.txt newname.txt renames a file only if newname.txt doesn’t already exist as a directory; if it does, Linux moves original.txt inside that directory instead of renaming it. Neither cp nor mv asks before overwriting an existing file by default. The classic guard is -i (prompt first), but the coreutils manual marks -n/–no-clobber as deprecated in favor of --update=none, which skips existing files without a prompt, or --update=none-fail, which does the same but reports an error for anything skipped; the Arch Linux packaged man page confirms both are current options.

Bulk renaming with the rename command

The single-file tools above don’t scale to hundreds of files at once, which is where the separate rename command comes in, and where most tutorials quietly assume a distro they may not be running on.

Which rename do you have installed?

Linuxize documents two unrelated programs sharing the name rename: a Perl-based version using regex substitution (rename 's/pattern/replacement/' files), and a C version from util-linux using plain string substitution (rename oldstring newstring files). Check which one you have with rename --version. On Debian and Ubuntu, Putorius confirms /usr/bin/rename typically resolves through /etc/alternatives/rename to the Perl file-rename script; on Fedora, RHEL, and Arch, the preinstalled rename is the util-linux version, and the Perl syntax requires installing a separate package (prename on Fedora, perl-rename on Arch). For a filename that starts with a dash, add -- before it so the shell doesn’t parse it as an option, and wrap Perl rename patterns in single quotes so the shell doesn’t expand them first.

• No error, no result: the command runs, prints nothing, and nothing changes, meaning the pattern matched nothing under the installed implementation.
• A syntax complaint about an unterminated expression: a sign you’re feeding Perl regex syntax to util-linux’s rename, which expects a plain old-string, new-string pair instead.
• Files renamed to literal garbage, slashes included: the reverse problem, a plain string pair fed to the Perl version, which reads it as a broken regex.

Running a regex-based rename one-liner from a tutorial on a system where rename belongs to util-linux does not produce an error describing regex syntax; it just fails to match anything or renames files to the literal pattern text, which is why the mismatch is easy to miss until the files are already wrong.

Why did the rename command I copied from a tutorial fail on my system?Most likely a mismatch between the Perl-based rename and the util-linux rename, which use completely different argument syntax; run rename –version to check which one is installed.

Preserving permissions and timestamps when copying

cp by default gives the copy the current date and time as its modification timestamp, not the original file’s metadata. Add -p to keep the original owner, permissions, and timestamps, or -a for a fuller copy that also preserves links and directory structure recursively. Neither flag matters for a plain mv, since renaming changes no metadata at all.

What happens when the copy or move crosses filesystems

Moving a file with mv across two different filesystems, such as from a home partition to a USB drive, doesn’t use the instant-rename trick at all: GNU coreutils documents that mv falls back to copying the data with the same routine cp -a uses, then deletes the source only after that copy succeeds. If that copy is interrupted partway through a multi-file move, the documented behavior leaves whatever already finished on the destination and whatever hadn’t started yet on the source, so a directory being moved between drives can end up split across both locations if the process is killed midway.

What happens if I move a file between two different drives?Its data is copied to the new drive first, and the original is deleted only after that copy finishes, so a cross-drive move briefly uses double the disk space and takes as long as a full copy.

Common mistakes when copying and renaming files

The five mistakes below share a common cause: treating cp and mv as safer, undo-friendly tools when neither offers a trash bin or confirmation by default.