How to find one email in a Google Takeout or mbox export

An .mbox is one plain text file with every message in it, one after another. Google Takeout exports Gmail this way, Thunderbird stores each folder this way, and most mail programs can export to it. To find one message in it, you don't need to import thousands of others into a mail program first.

Read it in place

mutt -R -f "All mail Including Spam and Trash.mbox"

For a scripted search, Python's standard library reads mbox files with no install. Save this as find.py:

import sys, mailbox, email, email.policy

box = mailbox.mbox(sys.argv[1], factory=lambda f: email.message_from_binary_file(f, policy=email.policy.default))
word = sys.argv[2].lower()
for key, msg in box.iteritems():
    if word in f"{msg['from']} {msg['subject']}".lower():
        print(key, msg['date'], msg['from'], msg['subject'], sep='  |  ')

python3 find.py "All mail Including Spam and Trash.mbox" lease

1  |  Thu, 04 Sep 2025 09:15:00 +0000  |  Priya Raman <[email protected]>  |  Re: Lease renewal for the flat
8  |  Thu, 18 Sep 2025 16:15:00 +0000  |  Priya Raman <[email protected]>  |  Lease renewal for the flat

The first number is the message's place in the mailbox. Write that one out as an .eml:

python3 -c 'import mailbox, sys; sys.stdout.buffer.write(mailbox.mbox(sys.argv[1]).get_bytes(int(sys.argv[2])))' "All mail Including Spam and Trash.mbox" 8 > lease.eml

get_bytes leaves out the mbox From line, so lease.eml is the message as it was sent, and any mail program opens it. Encoded subjects come out readable: the script prints Café Lumière where the file says =?UTF-8?Q?Caf=C3=A9....

Where the mailbox is in a Takeout download

Takeout sends an archive (.zip or .tgz), not the mailbox itself. Gmail's part is in a Mail folder inside it, usually as one file called All mail Including Spam and Trash.mbox. Google's own steps for requesting the export are in Download your Google data.

Unpack the archive before you open anything. The MBOX Viewer reads the bytes, not the file name, and says so if you give it the ZIP.

Gmail labels are a header

An mbox has no folders, so Takeout writes each message's labels into a header:

X-GM-THRID: 1810000000000063352
X-Gmail-Labels: Inbox,Home

Labels are separated by commas, and a message can have several. To see which labels a mailbox uses and how often:

import sys, mailbox
from collections import Counter

count = Counter()
for msg in mailbox.mbox(sys.argv[1]):
    for label in (msg['X-Gmail-Labels'] or '').split(','):
        if label.strip():
            count[label.strip()] += 1
for label, n in count.most_common():
    print(f'{n:6}  {label}')

    10  Inbox
     2  Receipts
     2  Important
     2  Home
     2  Sent
     2  Starred
     2  Category Updates
     1  Category Promotions
     1  Archived
     1  Spam

The MBOX Viewer reads the same header and shows the twelve most used labels as chips above the list. Pick one to see only those messages.

Why counting "From " lines gets it wrong

Each message starts with a line beginning From (no colon). The quick count is grep:

grep -c '^From ' old-client.mbox

5

That mailbox holds 4 messages. One body has a sentence that starts with "From ", written by a mail program that never escaped it as >From . Python's mailbox splits there too, and reports 5:

python3 -c 'import mailbox, sys; print(len(mailbox.mbox(sys.argv[1])))' old-client.mbox

5

The MBOX Viewer only counts a From line that has a sender and a date and follows an empty line, so it finds 4, and so does formail from procmail. On a clean export the counts agree: all of them found 14 in the Takeout file above.

Apple Mail and Thunderbird import it

Both can open an mbox, but by copying it into their own store. In Apple Mail it is File › Import Mailboxes; in Thunderbird, the ImportExportTools NG add-on. For a multi-gigabyte export that is a long wait and a second copy on disk, and the messages then sit in your mail program. Do that when you want the archive kept there. To find one message, read the file in place instead. We did not run these two imports for this guide; check the app's own help for your version.

How big a mailbox the browser handles

The MBOX Viewer opens files up to 16 GB. It reads the mailbox as a stream and keeps only a short summary per message (sender, subject, date, labels, where it sits in the file); a message is read from disk when you open it. Memory grows with the number of messages, not the size of the file. Run over a 281 MB test mailbox of 560,000 small messages, the same indexing code took 7 seconds and about 630 MB of memory, roughly 1 KB per message. A Takeout export of a few gigabytes with large attachments has far fewer messages than that.

A file of several gigabytes takes a while to read. The list fills as it goes, and you can search and open messages before it finishes.

Save the message

To keep a set of messages as separate files, narrow the list with search or a label first, then use the converter. It saves every message shown, one .eml per message, in one ZIP.

An .eml opens in Apple Mail, Thunderbird and Outlook. On a machine with none of them, open it in the Email Viewer.